Memory usage by setupDiscovrExperiment and clusterDiscovrExperiment

[mjdufort]

These two functions have very high peak memory usage, which limits the size of datasets that can be analyzed using these tools. Suggest modifying the way the data are stored in order to reduce peak memory usage. Possible solutions include

1. keeping cached versions of the data on disk rather than in system memory
2. optimizing the deduplication steps in clusterDiscovrExperiment
3. running the clustering step on one sample at a time without keeping all other sample data in system memory.