[unsupervised AI] Pin nightly Bokeh builds to Node 24

[YusefSyed]

Warning

This PR was written autonomously by an AI agent and has not been reviewed
by a human yet. Maintainers should ignore it until the human author has reviewed,
understood, and approved everything that the AI agent wrote.

The nightly environment currently fails while building Bokeh from source, before running tests. After pixi update -e nightly, Node reports v26.8.0-alpha.0.0.0; Bokeh's semver.satisfies(process.version, ">=16.0") check rejects that prerelease. This occurred on both the feature branch in #9358 and the documentation-only branch in #9359 (failed setup).

Constrain the nightly build tool to Node 24, matching Bokeh's own CI, and regenerate the lock with Pixi 0.72.1, the version used by the Upstream workflow. The resolved Node version is 24.19.0 on all four platforms.

Only the nightly environment changes in the lock. The solve also updates c-ares/libsqlite and removes libabseil on Unix platforms, and refreshes h5py, pandas, PyArrow, and SciPy nightly wheels. Retaining the previous pandas wheel was not solvable because its version is no longer available in the nightly index. Other environments and the locked Bokeh revision are unchanged.

Validation on macOS ARM64:

• pixi lock --dry-run --check --no-config: lock would not change.
• pixi install -e nightly --locked: successful with Node 24.19.0 and Python 3.14.6.
• Built a wheel from the exact Bokeh revision that failed in CI, d224d3ad29275d9d042c74384a3870f35f0b3394, using that environment's Node and Python. The build completed and the wheel contains the compiled BokehJS assets. This was a separate source build; the locked environment retains an older Bokeh revision.
• Parsed the lock and confirmed that all non-nightly environments are unchanged; git diff --check passed.

The full nightly test suite was not run locally. In particular, this does not address the separate SciPy sparse-matrix warning failures tracked in #9353.

[github-actions]

Unit Test Results

See test report for an extended history of previous test failures. This is useful for diagnosing flaky tests.

    40 files  ±0      40 suites  ±0   14h 23m 43s ⏱️ + 4m 45s
 4 160 tests ±0   3 981 ✅ +1    178 💤  - 1  1 ❌ ±0 
80 961 runs   - 1  76 719 ✅ ±0  4 241 💤  - 1  1 ❌ ±0 

For more details on these failures, see this check.

Results for commit e11d26f. ± Comparison against base commit dc182bd.