Improve fragment search query checks

[YukiMatsuzawa]

This PR implements the highest-impact, lowest-effort performance fix identified in the MSDIAL5 scan: replace two redundant Count() scans with a single All() check in FragmentSearcher.

The repo scan also surfaced 9 additional performance improvement candidates for follow-up:

1. MztabFormatExport.cs — replace .ToList()[0] with First(...).
2. AlignmentSpotProperty.cs — cache MspBasedMatchResult lookup.
3. AnalysisFileBeanModelCollection.cs — avoid Distinct().Count() uniqueness validation.
4. DataAccess.cs — replace Count() == 0 / repeated scans with cheaper checks.
5. PeakFilterViewModel.cs — cache repeated computed filter properties.
6. AbundanceRatioCalculator.cs — reduce repeated matrix allocation.
7. AlignmentSpotProperty.cs — cache AlignedPeakProperties task result access.
8. SpectraGroupingModel.cs / AlignmentCSVExporter.cs — reduce LINQ chain overhead on large collections.
9. DataAccess.cs — collapse multiple SWATH detection passes into a single scan.