fix(test): update exec-default and timeout tests to expect skip-on-failure

[zeroedin]

Summary

Fixes #168

Two SSR tests still expected fatal errors after #166/#167 aligned the other SSR tests to error collection. All SSR failures are per-page — page is skipped, original HTML preserved, build continues.

• "defaults to exec mode" — assert original HTML preserved instead of fatal error. Still proves exec mode is used (not stream) because the page is in the result.
• "enforces ssrCfg.Timeout" — assert page skipped with original HTML instead of fatal error. Timeout kills the command but the build continues.

Test plan

• No remaining Expect(err).To(HaveOccurred()) for SSR command failures in build_test.go
• All SSR tests consistently expect: no fatal error, original HTML preserved

🤖 Generated with Claude Code

[zeroedin]

PR #170 Review — Align Exec-Default and Timeout Tests with Skip-on-Failure Model

Scope: internal/pipeline/build_test.go (+21/-13). Updates two remaining tests that still expected fatal errors from BuildPhase2.

Changes

1. "BuildPhase2 defaults to exec mode when mode is empty"

Old: Expect(err).To(HaveOccurred()) + error message contains "nonexistent-ssr-tool" / "exec" / "not found"

New: Expect(err).NotTo(HaveOccurred()) + result has key + original HTML preserved with ds-card

The test's purpose shifts from "prove exec mode by checking the error message" to "prove exec mode by checking the page survived with original HTML." The new approach still proves exec mode was used (not stream) — if stream mode were used, NewStreamRenderer would fail differently (at process startup, before any page processing).

2. "BuildPhase2 exec mode enforces ssrCfg.Timeout"

Old: Expect(err).To(HaveOccurred()) — timeout kills process, build aborts with error

New: Expect(err).NotTo(HaveOccurred()) + timed-out page preserved with original HTML

Consistent with #164/#167: timeout is a per-page failure, not a build abort. The page that timed out gets its original HTML preserved; the build continues.

Completeness Check

With PRs #167 + #169 + #170, every BuildPhase2 test with a failing/nonexistent SSR command now uses the skip-on-failure model:

Test	Old Expectation	New Expectation
Phase 2 invokes command per page	fatal error	skip, preserve HTML
Command not found	fatal error	skip, preserve HTML
No local DSD fallback	conditional	unconditional, preserve HTML
Exec-default mode (#170)	fatal error	skip, preserve HTML
Timeout (#170)	fatal error	skip, preserve HTML
Skip failed pages (#164)	unconditional	unchanged
Preserve original HTML (#164)	unconditional	unchanged

All seven tests now agree: SSR failure = per-page skip + original HTML preserved + no build abort.

Verdict

Correct. Completes the error model alignment started in #167. No remaining tests expect fatal errors from SSR command failures. No issues.