docs(comments): drop comments that carry nothing the code does not

[Ilyes512]

Top of the stack on #135. Comments only — no behaviour changes, so no test or doc updates accompany it.

The repo's convention is a comment that explains why, and most of them earn their place. This prunes the ones that do not, across the stack's additions and the code that was already there.

Diff narration. templateRow explaining that Name "was a JSON key only because it was also a heading", and executeTemplate naming the Phase 7 and Phase 8 that introduced it. That belongs in a commit message; in the code the next reader has no idea what "was" refers to.

Restatement. // Check file content. above os.ReadFile, // Build dependency graph among computed keys. above the loop that builds it, the numbered 1–5 walkthrough of Execute's walk function, // validateGitHubName checks that a GitHub owner or repo name is valid., and three TestX verifies that… comments that only spell out their own test name. The two "copy so the caller's map is not mutated" notes in context.go say what the function doc two paragraphs up already promises.

Duplication. The rationale for splitting display strings from row values was written out four times, once per site the stack touched. It stays where the split is defined — Column and TableData — and the other sites point at it rather than restating it, so rewording it later is one edit instead of four. Same for the CI-hang rationale, which now lives on prompter.cannotPrompt alone.

Where a comment mixed a real constraint into a restatement, the constraint is kept and the restatement dropped: hasEmptySegment's "conditional directory exclusion" becomes the subtree exclusion it actually means, and validate.go keeps the non-obvious half about Template.Referenced already covering computed expressions.

Checks

task test, go test -race, go test -tags=integration -race ./internal/cmd/... and go build all pass.

One note: the local task lint reports 6 pre-existing SA5011 staticcheck findings in internal/cmd/metadata_test.go, a file this PR does not touch. They reproduce identically on #135's head and CI's golangci-lint is green there, so it is local toolchain drift rather than anything introduced here.

---

_{Stack created with GitHub Stacks CLI • Give Feedback 💬}