Implement Reader::iter_chromatograms (TIC chromatogram)

[Nabejo]

Fixes #21

Summary

Implements Reader::iter_chromatograms as a SpectrumSource override in crates/opensxraw/src/reader.rs, emitting exactly one total ion current chromatogram (MS:1000235 "total ion current chromatogram") built entirely from the already-decoded idx_records:

• one point per IdxRecord
• time_sec = retention_time_min * 60.0
• intensity = the record's Idx tic (cps)
• precursor_mz / product_mz = None

This is a pure wire-up of existing decoded fields into openmassspec_core::ChromatogramRecord; no new raw-format decode work is involved. As a result, TIC chromatograms now appear in the <chromatogramList> of the mzML OpenSXRaw produces, instead of falling through to the empty default impl.

Why the per-spectrum TIC reasoning does not apply here

iter_spectra deliberately leaves SpectrumRecord.total_ion_current as None because the Idx TIC is in cps (physically calibrated) and would fail the conformance suite's rel_close check against sum(raw intensities). That constraint is specific to the per-spectrum field. A TIC chromatogram is a separate trace of (retention time, cps) points; assert_source_invariants never inspects iter_chromatograms, so the mismatch does not apply. The per-spectrum field is left None exactly as before, so existing conformance behaviour is unchanged.

Scope: TIC only

BPC and SRM/MRM are intentionally left out, as the issue notes they need materially different net-new decode work:

• BPC would require a per-scan base peak, which nothing decodes today (base_peak_mz/base_peak_intensity are hardcoded None).
• SRM/MRM would require Q1/Q3 transition-level data, which is not decoded anywhere in this reader.

Suggest tracking both as separate follow-up issues if pursued.

Testing

• cargo fmt --all -- --check - clean
• cargo clippy --all-targets -- -D warnings - clean
• cargo test --workspace - all pass

Added:

• Two unit tests in reader.rs (synthetic idx_records, no corpus needed): verify a single TIC record with the correct CV term, None precursor/product, and that time_sec/intensity derive from retention_time_min/tic (including a large-cps regression guard confirming the value passes through verbatim rather than being tied to the per-spectrum field).
• One corpus-gated integration test in tests/conformance.rs (skips cleanly when the out-of-tree corpus is absent) asserting exactly one TIC chromatogram with one point per Idx record.

The existing conformance suite continues to pass unchanged.

Clean-room note

TIC values come only from bytes already parsed from the .wiff container by this project's own code. No comparison against Analyst / SCIEX OS / MultiQuant / msconvert output was made or used in verifying this feature, per CONTRIBUTING.md.

(Contributed by @Nabejo.)