fix(telemetry): drop profile field [skip-runtime-e2e]

[saurabhjain1592]

Summary

Drops the v1 telemetry profile field that was added in v8.0.0 (#161 / commit 15a448e). The AXONFLOW_PROFILE env var was already in use by the agent for governance enforcement (allowlist dev|default|strict|compliance per ADR-036), so emitting telemetry from a customer's correctly-configured governance profile (e.g. strict) would have produced HTTP 400 silent drops on the heartbeat path — the v1 telemetry validator accepts only dev|prod|unknown.

Refs getaxonflow/axonflow-enterprise#2033.

Why drop instead of rename

1. Env-var collision. AXONFLOW_PROFILE is owned by the governance-enforcement layer (platform/agent/profile.go, EnvProfile = "AXONFLOW_PROFILE") and has been since ADR-036. The v1 telemetry schema (#2004) picked the same name without grepping; the values do not overlap (strict|compliance vs dev|prod|unknown), so any customer who set the env var for governance would have started emitting 400-rejected pings.
2. Zero analytics consumers. A grep across platform/, ee/, and the analytics workflows for Profile.S | .Profile\b | profile.*dimension (excluding _test.go and the checkpoint-service) returns nothing — no aggregator, no daily-report dimension, no dashboard reads the field. Removing it loses no signal in use today.
3. deployment_mode already covers the topology dimension the field was meant to add: self_hosted | community_saas | unknown, single canonical form, no env-var collision, has analytics callers. Untouched by this PR.

Renaming would force a 9-repo coordinated bump just to allowlist-split a value nothing currently reads. Drop is the cleanest fix.

Validator-vs-caller audit

$ grep -rn 'AXONFLOW_PROFILE\|Profile\b' --include='*.go' .
(no output)

Verified zero AXONFLOW_PROFILE / Profile references remain across the 4 changed files. find . -name 'baseline*' returned testdata/wire_shape_baseline.json — grep -i 'profile' against it: empty.

Hunk-by-hunk self-review (HARD RULE #1)

For each of the 4 hunks, the 5-question protocol from feedback_self_review_is_mandatory_every_pr.md:

Hunk 1 — telemetry.go struct field removal (3 lines doc + 1 line field).

1. Does this match the intent in the PR title? Yes — drops the Profile field from telemetryPayload.
2. Are there any references that are now broken? grep confirms no — the only callers were the assignment site (Hunk 3) and tests (Hunks 5-6).
3. Backward-compat consideration? Server-side ignores unknown JSON fields per json.Unmarshal semantics; no client crashes from old-server-vs-new-client. Reverse direction (new server, old client emitting profile) is also fine — server will validate-and-drop or accept under the JSON unmarshaller.
4. Anything load-bearing about the doc comment? No — purely descriptive, references the field that no longer exists.
5. Does this leave any dead code? No — Hunks 2-3 remove the read + assignment.

Hunk 2 — telemetry.go os.Getenv("AXONFLOW_PROFILE") read removal (8 lines).

1. Intent match? Yes — removes the env-var read that was the collision source.
2. Broken refs? profile local var was used only in Hunk 3.
3. Imports unused? strings still used (TrimPrefix on runtime.Version, EqualFold + TrimSpace + ToLower elsewhere); os still used (Getenv for AXONFLOW_TELEMETRY etc). No orphan imports — go build confirms.
4. Side effects? None — pure read of env, no caller behavior change beyond not setting that one struct field.
5. Tests still cover the surrounding paths? Yes — TestSendTelemetryPing_Success + TestBuildPayload deployment_mode + stream subtests still run.

Hunk 3 — telemetry.go payload literal Profile: profile removal (1 line).

1. Intent match? Yes — removes the payload-side use site.
2. Compiles? go build ./... passes.
3. Tests? go test ./... passes 7.6s.
4. Field ordering / struct literal validity? Unchanged — Go allows omitting any field in a struct literal.
5. Any other emission site? No — grep -rn 'telemetryPayload{' --include='*.go' returns just this one (and tests that decode into the struct).

Hunk 4 — telemetry_test.go removal of received.Profile assertion in TestSendTelemetryPing_Success (3 lines).

1. Intent match? Yes — drops the assertion for the removed field.
2. Test still meaningful? Yes — TestSendTelemetryPing_Success retains assertions on TelemetryType, DeploymentMode, OS, Arch, RuntimeVersion, Features, InstanceID. Coverage of the heartbeat happy-path remains.
3. Other call sites for the same assertion? Hunks 5-6 cover the dedicated subtest.
4. Helper utilities deleted? None — pure assertion removal.
5. Race / setup change? None — same t.Setenv setup.

Hunks 5-6 — telemetry_test.go removal of t.Run("profile from AXONFLOW_PROFILE; unknown when unset", ...) subtest (33 lines).

1. Intent match? Yes — drops the dedicated profile-sourcing subtest entirely.
2. Other coverage? TestBuildPayload retains its deployment_mode classifies from endpoint host and stream=sandbox tag emitted only for sandbox mode subtests; both run independently of the dropped subtest.
3. Compiles? go test -c ./... passes.
4. No shared state? The dropped subtest used only t.Setenv (per-subtest scoped); no fixture cleanup needed.
5. Closing brace alignment? Verified — the next subtest (stream=sandbox) is indented correctly.

Test results

$ go build ./... && go vet ./... && go test ./...
ok  	github.com/getaxonflow/axonflow-sdk-go/v8	7.642s
ok  	github.com/getaxonflow/axonflow-sdk-go/v8/interceptors	12.335s
ok  	github.com/getaxonflow/axonflow-sdk-go/v8/internal/wireshape	0.514s

All green. Wire-shape baseline (testdata/wire_shape_baseline.json) had no profile reference, so no baseline refresh needed.

Migration notes

• Customers who set AXONFLOW_PROFILE for governance enforcement on the agent (the original ADR-036 use case): no change. The env var is read by the agent for that purpose unchanged.
• Customers who set AXONFLOW_PROFILE solely to influence telemetry: no longer needed — the SDK no longer reads it for telemetry. Telemetry's topology dimension lives on deployment_mode (auto-derived from endpoint host).

DO NOT TAG / PUBLISH

This is part of a coordinated train (9 client PRs + server PR per the session-2033 brief). Tagging is operator-gated and happens only after the full train merges. Do not tag v8.0.1 from this PR.

Test plan

• go build ./...
• go vet ./...
• go test ./... — all green
• grep -rn 'AXONFLOW_PROFILE\|Profile\b' --include='*.go' . — zero hits
• CI green
• Coordinated merge with the rest of the session-2033 train
• Runtime proof against staging-checkpoint.getaxonflow.com (cross-train step, not gated by this PR alone)

Skip-runtime-e2e justification

This PR is part of the #2033 coordinated train across 1 server + 9 client repos. Per the session-2033 brief, runtime proof is deferred to the post-server-merge staging-checkpoint deploy:

1. axonflow-enterprise#2035 (server) merges + gh workflow run deploy-checkpoint.yml -f environment=staging deploys the new code to staging-checkpoint.getaxonflow.com.
2. Each of the 9 client builds is then driven against staging-checkpoint with verification that (a) POST returns 200 (no validator complaint about missing profile) and (b) the resulting DDB row has no profile attribute.
3. EVIDENCE lands at runtime-e2e/profile_field_removal/EVIDENCE/<utc-ts>/ post-deploy.

Adding a same-PR runtime-e2e/ test here would either:

• exercise a fake checkpoint server (forbidden by lint-no-mocks-in-runtime-e2e.sh), or
• run against the live deployed Lambda BEFORE the server PR has deployed, which would either silently pass (ignored field) or fail (depending on deploy ordering) — neither outcome is informative.

The post-deploy proof against staging-checkpoint is the only meaningful runtime test for this train.