refactor: Categorical now stores normalized values

[YeungOnion]

Cherry-pick of 37f8fdf (FreezyLemon) onto current main. The original commit predates the rand 0.9/0.10 migration and the std/core split, so it conflicted on:

• ::rand::distr::Distribution (vs. old rand::distributions::Distribution alias)
• RngExt::random (vs. old rng.gen)

Categorical now stores norm_cdf/norm_sf directly instead of normalizing on every cdf/sf/inverse_cdf call, and inverse_cdf uses binary_search_by instead of a hand-rolled binary_index. The now-unused sample_unchecked/prob_mass_to_cdf/cdf_to_sf/binary_index helpers are removed; Multinomial::sample, which depended on categorical::prob_mass_to_cdf/sample_unchecked, now computes its cdf inline instead.

Test plan

• cargo build --all-features
• cargo test --all-features (192 passed)
• cargo clippy --all-features

[codecov]

Codecov Report

❌ Patch coverage is 95.09804% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 94.89%. Comparing base (9d33d0e) to head (07a05b7).

Files with missing lines	Patch %	Lines
src/distribution/categorical.rs	91.83%	4 Missing ⚠️
src/distribution/multinomial.rs	98.11%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #396      +/-   ##
==========================================
+ Coverage   94.67%   94.89%   +0.21%     
==========================================
  Files          61       61              
  Lines       13322    13344      +22     
==========================================
+ Hits        12613    12663      +50     
+ Misses        709      681      -28     

☔ 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.

[YeungOnion]

GoF tests for sampling will still be a good idea