Revert narrative review v1 (PRs #52-56)

[neuromechanist]

Summary

Revert the entire v1 narrative-review epic (PRs #52, #53, #54, #55, #56) so v2 can be built with proper skill invocations.

Why: v1 was produced by bypassing /manuscript:manuscript-formatting and /manuscript:manuscript-writing and treating /figures:svg-figure as the entire figure pipeline instead of using /figures:scientific-figure as the composer. This was contrary to the explicit instruction to use those skills, and the figures were not publication-ready (figure-qa findings deferred). This PR removes v1 from main so v2 can replace it cleanly.

Reverts:

• 7045be3 Phase 5: final assembly + icon polish (Phase 5: final assembly + icon polish #56)
• 3fda6dd narrative review phase 4: paper-review + humanizer (Phase 4: paper-review + humanizer #55)
• 0b80fc9 narrative review phase 3: full manuscript draft (Phase 3: full manuscript draft #54)
• e200f6d narrative review phase 2: 4 figures + QA (Phase 2: narrative review figures #53)
• 808dd6a narrative review phase 1: TiCS scaffold (Phase 1: TiCS scaffold for narrative review #52)

Five git revert commits applied in reverse order on a clean branch off main. Net effect: manuscript/narrative-review/ is removed.

Test plan

• Five revert commits apply cleanly with no conflicts
• manuscript/ directory is gone after revert chain
• No other content in the repo touched the manuscript/ path before v1
• markdownlint and typo pre-commit hooks pass on the revert branch