Cache ingested scratch DB

[neilotoole]

Is your feature request related to a problem? Please describe.

Consider a non-SQL source, such as an Excel file. When sq accesses that source (e.g. sq inspect @excel), it ingests the data into the scratch DB on each command. For large files, this can take some time... seconds, or even minutes. If the file doesn't change between commands, this is wasted work.

Describe the solution you'd like

sq could ingest the data the first time, e.g. into a SQLite scratch DB in /tmp. On the next command, if the source file hasn't changed, sq could skip the ingest step, and use the already populated scratch DB.

[neilotoole]

This is implemented in sq v0.47.0.