test: strengthen P1 effect regression assertions (#146, #162)

[sonupreetam]

Summary

• test: strengthen MapMutation/SliceMutation regression test assertions #146: MapMutation/SliceMutation tests now assert exact effect count (1) and correct target name ("m"/"s") instead of boolean presence checks. Applied at both the unit layer (p1effects_test.go) and integration layer (analysis_test.go).
• test: add edge case coverage for hasHTTPResponseWriterMethods and matchesWriteSignature #162: Added edge case fixtures and tests for hasHTTPResponseWriterMethods / matchesWriteSignature:
  • PartialResponseWriter (Write + Header, no WriteHeader) — verifies WriterOutput is emitted but HTTPResponseWrite is not.
  • WrongWriteReturn (Write returns int, not (int, error)) — verifies neither WriterOutput nor HTTPResponseWrite is emitted.

Test plan

• go test -race -count=1 -short ./internal/analysis/ passes
• All new assertions pass against existing implementation
• CI green

Refs #146, #162

[yvonnedevlinrh]

LGTM - the count + target pattern is a real improvement over boolean presence checks.

Two observations:

1. TestAnalyzeP1Effects_WrongWriteReturn — vacuously true assertions: HandleWrongWriteReturn produces zero effects, so both hasEffect checks pass trivially (an empty slice has no effects of any type). Consider adding if len(effects) != 0 { t.Errorf("expected no effects for WrongWriteReturn, got %d", len(effects)) } to explicitly assert the empty result. Without it, a future change that accidentally introduces a different effect type (e.g., some new call-expression effect) would pass undetected.

2. Partial #162 coverage: This PR covers the partial method set case (PartialResponseWriter) and one near-miss Write signature (Write([]byte) int), but #162 also lists three other near-miss cases (Write(string), Write([]byte) (int64, error), Write([]byte, int)) and suggests an internal unit test for matchesWriteSignature directly. If #162 gets closed after this PR merges, those cases would be silently dropped. Might be worth either adding them here (they're ~5-line fixture types each) or noting in #162 which items remain open.

[sonupreetam]

@yvonnedevlinrh Thank you for catching this. I did the following:

• Fix observation 1 in this PR to add len(effects) != 0 guard to TestAnalyzeP1Effects_WrongWriteReturn.
• Commented on #162 documenting what this PR covers and what remains open (3 near-miss signatures + internal unit test).
• Created PR test: add remaining matchesWriteSignature near-miss coverage (#162) #196 in draft now, would need to rebase after this PR merge.

[yvonnedevlinrh]

LGTM