Add an option to tide-search called "--store-distributions" that is analogous to "--store-spectra." The resutling protocol buffer object would store, for each spectrum, the final column of the DP matrix computed, as well as an indicator of which bin corresponds to a score of 0. Note that if the spectrum is searched with a wide precursor tolerance, then multiple columns would need to be stored, one for each integral bin. Using this option would greatly speed up searching, e.g., in the context of cascade search.