Consider always validating incoming data in kcidb.Client.submit()

The submitters are told to use `kcidb.Client.submit()` to send their data, and that it's equivalent to `kcidb-submit`, but the former doesn't actually validate the supplied data (for performance reasons). Consider either always validating the data passed to that function, documenting that explicit validation is required, or adding a separate function that doesn't validate the data for those who need performance (like internal library users).

Start with the kcidb.Client.submit() entry point and compare its behavior with kcidb-submit, focusing on where incoming data validation is skipped. Determine which of the proposed behaviors is appropriate for submitters and internal library users, then document or implement the chosen contract and verify that valid and invalid submissions behave as intended.

Written by the indexing model from the issue text.