chore(release): prepare for 0.4.10-rc.2

[github-actions]

Prepares v0.4.10-rc.2, bumped by cargo release rc.

The version and the changelog are the whole diff. Merging this tags
v0.4.10-rc.2 and starts the release, which builds the binaries and
plays the strength match against the previous release.

Close it without merging to call the release off, nothing has been
tagged or published yet.

[github-actions]

Speed against 06c1abdb

Measured on AMD EPYC 7763 64-Core Processor.
Both sides built with rustc 1.99.0 (b940084d7 2026-09-28).

Both sides built and run on this runner in this job, the way
scripts/speed.sh measures a perf commit. Each round runs
both sides on a layout of its own, the same compiled code with
its code and data shuffled and moved, so the 95% interval carries
where the code landed as well as the run. The verdict holds the
interval against a 1% threshold. The default layout, the one a
release ships, is measured after as a diagnostic. The node
counts are the search's: they move when the search does, and a
speed change leaves them alone.

round     base nps  candidate nps  change
    1      4573516        4588972   +0.3%
    2      4552539        4535233   -0.4%
    3      4577753        4658059   +1.8%
    4      4553590        4602144   +1.1%
    5      4542575        4486765   -1.2%
    6      4660892        4589488   -1.5%
    7      4509012        4659940   +3.3%
    8      4566023        4517621   -1.1%
    9      4699790        4637207   -1.3%
   10      4668676        4636027   -0.7%
   11      4620325        4601061   -0.4%
   12      4573247        4604424   +0.7%
   13      4594286        4601903   +0.2%
   14      4453987        4577171   +2.8%
   15      4492113        4575920   +1.9%
   16      4519623        4689845   +3.8%
   17      4463379        4534513   +1.6%
   18      4483998        4596523   +2.5%
   19      4567850        4568561   +0.0%
   20      4621056        4641455   +0.4%
   21      4652912        4568811   -1.8%
   22      4611380        4559548   -1.1%
   23      4612067        4561997   -1.1%
   24      4639211        4593854   -1.0%
   25      4655004        4525205   -2.8%
   26      4623932        4704902   +1.8%
   27      4620427        4555527   -1.4%
   28      4610657        4645345   +0.8%
   29      4590819        4588161   -0.1%
   30      4529949        4570024   +0.9%
   31      4476000        4616460   +3.1%
   32      4594203        4662057   +1.5%
   33      4624359        4662102   +0.8%
   34      4577910        4614081   +0.8%
   35      4623950        4618903   -0.1%
   36      4534416        4599258   +1.4%
   37      4633300        4412716   -4.8%
   38      4613108        4556578   -1.2%
   39      4511582        4575159   +1.4%
   40      4645654        4567818   -1.7%

              nodes    time  median nps  faster half
base       11916992  2.59 s     4592511      4631259
candidate  11916992  2.60 s     4591671      4632278
change                            -0.0%        +0.0%

paired change +0.2%, 95% interval -0.3% to +0.8%
diagnostic, on the default layout alone -0.8%, 95% interval -1.9% to +0.2%, 13 rounds

no change beyond ±1.0%: the whole interval is inside it

Speed: +0.2% (bench nps, 95% interval -0.3% to +0.8%, 40 interleaved rounds over shuffled layouts vs 06c1abdb)

The instructions each side's bench executed, counted under
cachegrind. The count repeats to within a few hundred
instructions, so a small change here is a real one, but it
prices instructions only: cache misses, mispredicted branches
and where the code lands are the speed's to show.

instructions against 06c1abdb, one cachegrind run a side, bench

             instructions     nodes  per node
base       15,968,841,487  11916992    1340.0
candidate  15,968,841,530  11916992    1340.0
change             +0.00%              +0.00%

within ±0.7%, as far as edits not made for speed have moved the count