[query] Remove BlockMatrix persist from Backend interface

[ehigham]

This change is split out from a larger refactoring effort on the various Backend
implementations. The goals of this effort are to provide query-level
configuration to the backend that's currently tied to the lifetime of a backend,
reduce code duplication and reduce state duplication.

In this change, I'm removing blockmatrix persist/unpersist from the Backend
interface by adding BlockMatrixCache: mutable.Map[String, BlockMatrix] to
ExecuteContext. The various reader/writer implementations simply fetch the
block matrix from this cache. For the spark backend, this is backed by a cache
whose lifetime is tied to the spark backend. Since block matrices are not
supported in the local and service backends, the cache is an empty map.

Note that block matrix persist is broken in python (#14689)

[ehigham]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• [query] Lowering + Optimisation with implict timing context #14731 : 2 dependent PRs (#14732 , #14748 )
• [query] Lift backend state into {Service|Py4J}BackendApi #14698
• [query] Move LoweredTableReaderCoercer into ExecuteContext #14696
• [query] unify backend rpc #14693
• [query] Remove persistedIr from Backend interface #14692
• [query] Remove lookupOrCompileCachedFunction from Backend interface #14691
• [query] Remove BlockMatrix persist from Backend interface #14690 👈 (View in Graphite)
• [query] Expose references via ExecuteContext #14686
• [query] Unify CloudCredentials and Simplify BatchClient #14684 : 1 other dependent PR (#14747 )
• [query] Use sourcecode.Enclosing to handle timed blocks implicitly #14683
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.