perf(tl): vectorize ancestral_states sum for array data

[colganwi]

Summary

• Replaces per-feature tree traversal loop with a single bottom-up pass using numpy for the sum method
• Reduces complexity from O(n_nodes × n_features) tree traversals to one

Test plan

• Existing ancestral_states tests pass
• Verify correctness of sum output against previous implementation
• Benchmark on large arrays to confirm speedup

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.04%. Comparing base (2afbd61) to head (789fad2).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #56      +/-   ##
==========================================
+ Coverage   93.99%   94.04%   +0.04%     
==========================================
  Files          35       35              
  Lines        2983     3004      +21     
==========================================
+ Hits         2804     2825      +21     
  Misses        179      179              

Files with missing lines	Coverage Δ
src/pycea/tl/ancestral_states.py	97.46% <100.00%> (+0.24%)	⬆️

🚀 New features to boost your workflow:

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

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 789fad2729

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Preserve NaNs when summing array-valued descendants

Using np.nansum here changes the sum reconstruction semantics by treating missing child values as zero instead of propagating missingness. In the previous per-index implementation (and still in scalar sum), a child value of NaN made the parent NaN for that feature ([NaN, 2] -> NaN), but this branch now returns 2. That silently converts unknown measurements into observed totals and can materially change reconstructed ancestral states for arrays with partial missing data.

Useful? React with 👍 / 👎.