revert: charmbracelet group bump #636 (lipgloss + ultraviolet) — lingering selection output

[zimeg]

Changelog

Fix interactive selection prompts (e.g. the slack create category picker) leaving stale/lingering rendered lines on screen after a selection is made.

Summary

Reverts the charmbracelet group bump from #636 across both deps it changed. #636 was titled as a charm.land/lipgloss/v2 v2.0.5 → v2.0.6 bump, but its go.mod diff also carried github.com/charmbracelet/ultraviolet f5a850f9 → 006e29f9. The lingering-selection regression is in ultraviolet's terminal renderer, not lipgloss:

• slack create → select a category leaves residual lines on screen after the pick.
• The teardown/erase behavior is owned by ultraviolet (huh → bubbletea → ultraviolet's renderer), which was substantially rewritten between f5a850f9 and 006e29f9 (new line-drift/repaint handling and changed inline erase-height logic). The wide selector glyph exercises exactly that path.
• lipgloss v2.0.6 requires the new ultraviolet via MVS, so the two are coupled — pinning lipgloss alone left this repo resolving ultraviolet to 006e29f9, and the regressed renderer was never actually reverted. Both are pinned back together: lipgloss → v2.0.5, ultraviolet → f5a850f9.

An earlier revision of this PR pinned only lipgloss; that did not fix the symptom (verified on a real terminal — selections still lingered), which is what surfaced ultraviolet as the true cause.

Testing

• go build ./... passes with lipgloss v2.0.5 + ultraviolet f5a850f9.
• slack create → Automation App category selection verified by hand on a real terminal: prompt tears down cleanly, no lingering lines. The prior config (lipgloss v2.0.5, ultraviolet still 006e29f9) reproduced the lingering; pinning ultraviolet back to f5a850f9 resolves it.
• The lingering artifact only renders through the interactive inline renderer on a real TTY (accessible-mode / model-stepping tests bypass it), so it is verified by hand rather than in headless CI.

Notes

• The dependabot ignore rule fences both charm.land/lipgloss/v2 (>= 2.0.6) and github.com/charmbracelet/ultraviolet (>= 006e29f9) so the next charmbracelet group bump can't silently re-raise either half and undo the revert.
• This is an interim pin — a follow-up issue upstream will track the proper ultraviolet fix (the renderer's erase-sizing on lines with wide glyphs), so we can un-pin both deps together once it ships.

Requirements

• I've read and understood the Contributing Guidelines and have done my best effort to follow them.
• I've read and agree to the Code of Conduct.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 78.19%. Comparing base (9b4235c) to head (80fed66).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #669   +/-   ##
=======================================
  Coverage   78.19%   78.19%           
=======================================
  Files         239      239           
  Lines       18149    18149           
=======================================
  Hits        14192    14192           
  Misses       3957     3957           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[zimeg]

@mwbrooks @srtaalej Super appreciate the fast reviews! More testing finds that pinning the ultraviolet dependencies required from lipgloss keeps this change stable 💄 ✨

Let's merge this now but search for upstream fix to remove our @dependabot changes here since I don't believe we can unit test this with confidence at this immediate time.