MAST: support query result caching

It'd be great if MAST also supports query result caching.

One use case is from `lightkurve` package: https://github.com/KeplerGO/lightkurve/issues/461

For `lightkurve` use case, query result caching is needed at
`mast.Observations.query_criteria()` function.

Start by reading the MAST implementation of mast.Observations.query_criteria() and the linked lightkurve use case. Determine how query results should be cached for that entry point, then verify that repeated queries return cached results while preserving existing behavior.

Written by the indexing model from the issue text.