Properly handle joins and propagate invalidation assessment

[ahuang11]

I am now properly able to back out of the ephemeral table and consistently get the desired result.

The root issue was that if there were multiple tables within a single source that needed to be joined, the key a_source_name was overridden, so no joins really happened. sources[a_source_name] = (a_source, a_table)

I refactored it so the table name is the key; and if there are duplicate it prefixes with the source name and then re-extracts it out when fetching the table.

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 0% with 56 lines in your changes missing coverage. Please review.

Project coverage is 56.51%. Comparing base (d698872) to head (910f939). Report is 1 commits behind head on json_to_yaml.

Files with missing lines	Patch %	Lines
lumen/ai/agents.py	0.00%	46 Missing :warning:
lumen/ai/assistant.py	0.00%	9 Missing :warning:
lumen/ai/models.py	0.00%	1 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## json_to_yaml #704 +/- ## ================================================ - Coverage 56.59% 56.51% -0.08% ================================================ Files 98 98 Lines 11698 11713 +15 ================================================ Hits 6620 6620 - Misses 5078 5093 +15 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.