Stream predicted libraries and calculate isotope envelopes on demand

[jspaezp]

Summary

• Push each predicted spectrum into the target columns builder as it arrives; retain only the final arena and fragment intensity sidecar.
• Calculate precursor isotope envelopes when scoring requests them. Composition mode stores four bytes of C/S counts per row; mass mode uses stored geometry.
• Keep library-wide isotope method selection and numerical range validation. Log when finalization starts.
• Track sink callbacks with explicit AwaitingHeader / Building / Finished states. Carry typed LibraryProvenance through finalization and convert to JSON only when recording the search config.
• Default FASTA prediction to the 12 strongest fragments per precursor after intensity filtering. Materialize effective prediction settings before the initial configuration log and config_used.json write.

Validation

• task fmt
• cargo test -p timsseek_cli (110 unit tests and 1 integration test passed)
• cargo test -p timsseek isotope_plan (7 passed)
• cargo clippy -p timsseek -p timsseek_cli --all-targets -- -D warnings
• CLI smoke check: the initial resolved-config log shows digestion settings and max_fragments: 12 before input validation.

No full-scale runtime benchmark yet.

[github-actions]

Apex-finder bench

cargo run -p apex_sim --release --example bench -- 1000 2

Sensitivity + timing across canonical scenarios

=== summary (n=1000, tol=±2 cycles) ===
  scenario                      pass2%  pass1%   medErr    us/run
  clean                          100.0   100.0        0     27.04
  moderate_noise                  92.7    92.7        0     26.65
  high_noise+interference         59.3    59.3        1     26.49
  heavy_interference               7.4     7.4       88     26.45
  mismatched_library              59.8    59.8        1     26.48
  absent_top_fragment             16.0    16.0       65     26.48
  absent_precursor                59.3    59.3        1     26.57

=== broad apex-finding (n=500, tol=±2 cycles) ===
  scenario                      pass2%  pass1%   medErr    us/run
  broad_clean                    100.0   100.0        0    168.87
  broad_moderate_noise            50.6    50.6        1    165.83
  broad_high_noise+interf         27.6    27.6      308    165.93
  broad_hard_3x_density            0.8     0.8      487    165.67
  broad_mismatched_library        41.0    41.0      161    165.76
  broad_measured_density          49.6    49.6       13    193.71

=== narrow recovery (n=1000, tol=±2 cycles) ===
  scenario                      pass2%  pass1%   medErr    us/run
  narrow_clean                   100.0   100.0        0     17.91
  narrow_moderate_noise           81.3    81.3        0     17.56
  narrow_high_noise+interf        65.8    65.8        1     17.64
  narrow_hard_3x_density          10.9    10.9       34     17.42
  narrow_mismatched_library       62.8    62.8        1     17.61
  narrow_measured_density         83.6    83.6        0     19.98

=== narrow score discrimination (AUC, n_seed_pairs=1000) ===
  scenario                         AUC   med+signal    med-noise
  narrow_clean                   1.000     4.329e14      7.199e7
  narrow_moderate_noise          0.870      4.054e9      3.054e8
  narrow_high_noise+interf       0.830      2.761e9      3.602e8
  narrow_hard_3x_density         0.684      3.644e9      1.244e9
  narrow_mismatched_library      0.842      2.172e9      3.575e8
  narrow_measured_density        0.858      4.341e9      2.562e8

commit edd3753