ci: add Rust CI workflow, cargo-deny and typos config (closes #19)

[Dani-giron]

Summary

The CI workflow was left over from an earlier Python version of the project (sets up Python 3.9, runs unittest) and did not build, test or lint any Rust code. It also did not trigger on the apotheosis2 branch.

Changes

• cargo fmt applied to hnsw.rs and record.rs (mechanical, no logic changes) and every clippy warning fixed across src/, verified one by one (details in individual commits). Both were prerequisites: a fmt/clippy CI job added on top of unformatted or warning-producing code would fail on day one.
• Cargo.toml: added rust-version = "1.88", the version that stabilizes let-chains, used by some of the clippy fixes.
• .github/workflows/ci.yml: replaced with a Rust pipeline (fmt, clippy, test, release build, cargo-deny, cross-platform check, typos, MSRV, docs), triggered on the apotheosis2 branch.
• deny.toml: license and advisory policy for cargo-deny.
• _typos.toml: exceptions for the two Spanish words in README's Funding support section (Estratégicos, Ciberseguridad), which the English dictionary would otherwise flag.

Test plan

• cargo fmt --check, cargo clippy --all-targets --all-features -- -D warnings, cargo test --all-features, cargo build --release --all-features and cargo doc --no-deps --all-features (with RUSTDOCFLAGS=-D warnings) all pass locally.
• cargo-deny and typos-cli could not be installed locally (unrelated toolchain issue); not verified locally, will be validated by the CI run itself on this PR.

Closes #19

[danielhuici]

Here we are modifying critical part of the code. Could you check that the construction and query times have not been incremented? This is, compare times before and after this commits

[Dani-giron]

Ran test_numbers (release build), two runs each before and after this
PR's clippy fixes (base commit 66455fd vs this branch).
Before: construction 3.358s / 3.458s, queries 145.95ms / 142.38ms
After: construction 3.197s / 3.322s, queries 141.96ms / 141.16ms
Matches: 5000/5000 in all four runs (correctness unchanged).
No regression: the before/after ranges overlap with the run-to-run variance seen within the same commit, if anything slightly faster after. Full terminal output in the next comment.

[Dani-giron]

Before (66455fd) - run 1:

HNSW Model created in: 3.3579866s
Matches: 5000/5000
Queries took: 145.9505ms

Before (66455fd) - run 2:

HNSW Model created in: 3.4579097s
Matches: 5000/5000
Queries took: 142.3787ms

After (this branch) - run 1:

HNSW Model created in: 3.1969212s
Matches: 5000/5000
Queries took: 141.9565ms

After (this branch) - run 2:

HNSW Model created in: 3.3216627s
Matches: 5000/5000
Queries took: 141.1579ms

[danielhuici]

Looks good to me