fix(scan-event): decode Exploris coefficients + MS2 precursor

[oskarsari]

Closes #2 .

What

Decodes two scan-event-derived values that were missing for Orbitrap Exploris files (v66, body_size 136): the frequency->m/z calibration coefficients (so the profile m/z converts correctly) and the MS2 precursor m/z (so dependent scans report their precursor and the rendered filter includes it).

Why

The v66 parser read the nparam/coefficients block at a fixed body_size - 64 and the MS2 reaction record at body offset 8. Both hold for Q Exactive (body_size 144) but not Exploris (136):

• On Exploris MS1, body_size - 64 landed on a zero word, so coefficients came back empty and Profile::to_mz_intensity passed the raw frequency through unconverted (profile m/z was wrong, e.g. ~600-2050 instead of ~40-470).
• On Exploris MS2, the reaction record starts at body offset 4 (no leading count word), so the offset-8 parse read the precursor double's low word as a reaction count, found it implausible, and returned no reaction. The precursor and HCD energy were lost and the filter dropped the precursor.

How

• Coefficients: read the nparam/coefficients block immediately after the located scan-window FractionCollector (offset 80 for the offset-64 FC family) instead of a fixed body_size - 64. On body_size 144 (Q Exactive) this is the same offset 80, so that path is unchanged; Exploris (136) now finds its A/B/C.
• Precursor: when an MS2+/dependent, non-tribrid scan yields no reaction from the offset-8 parse, fall back to the Exploris layout where the reaction record begins at body offset 4 (precursor f64 at body[4..12]).

Both changes are additive/guarded; other instrument families take the existing paths.

Testing

cargo fmt --check, cargo clippy --all-targets -- -D warnings, and cargo test --all pass. Validated against the vendor reader on Orbitrap Exploris 120 acquisitions:

• Profile m/z now converts correctly, matching the vendor reader to the last digit.
• MS2 precursor and activation energy match exactly, and build_filter now renders e.g. ... ms2 100.0757@hcd3.00 ....

No regression on Q Exactive Plus files (identical coefficients/profile before and after). All format knowledge is from public-corpus files and my own analysis.

Checklist

• CHANGELOG [Unreleased] updated
• cargo fmt / cargo clippy -D warnings / cargo test --all pass
• ASCII-only source