Storing raw data, so N*int32 in size.

Run length encoding, assumption is many numbers have long runs.

0's all have a count after them.

Enum value maps for Experiment_Location_SpectrumCompression.

Enum value maps for Experiment_Location_SpectrumCompression.

Enum value maps for Experiment_MetaDataType.

Enum value maps for Experiment_MetaDataType.

Enum value maps for Quantification_QuantDataType.

Enum value maps for Quantification_QuantDataType.

Aligned context image - describes which PMC images are defined for, and if they're trapezoid corrected.

Spectrum data, stored as specified in spectrum_compression.

Storage of pseudo-intensities.

Matched context image - list of context images which have been transformed to match a given PMC's MCC image (in aligned_context_images), for example WATSON images.

If we have pseudo-intensity data, this will contain the labels of the elements from the CSV, along with their channel start/end.

A long time after we had individual datasets as above, we decided to support combination of multiple scans into one dataset.