Raw entity ids support

[tomershefi]

This upgrade supports GTFS files that contain entities that reference other entities which do not exist in the GTFS.
A good example maybe a stop_time entity that references a trip id that does not exist in trips.txt.
This is a very common problem because many providers supply such GTFS files.
Up until now we overcame this issue by using extensions but this is very inefficient especially in large GTFS files.

Thank you very much!

[CLAassistant]

All committers have signed the CLA.

[sheldonabrown]

@tomershefi can you also add a test case that demonstrates how this works?

I have a concern for the case where clients of this module expect the referential integrated to be maintained and in fact enforced. Does this PR bypass that? If so, can it be configured optionally?

[leonardehrenfried]

So far OBA has been my first line of defense against non-compliant GTFS feeds and it would be lovely if the referential integrity guarantees could be maintained. I would like to get an exception about this as early as possible.

[tomershefi]

Hey, @sheldonabrown @leonardehrenfried
This does not interfere with the current referential integrity, and that is also why i did not add any tests. The invalid GTFS file would fail exactly as it did before.
Any code base that extends this lovely library can use this new feature as it sees fit or even not at all.

[tomershefi]

Hey, @sheldonabrown @leonardehrenfried.
Any updates?

[leonardehrenfried]

I frankly would not be in favour of supporting invalid gtfs files. I suggest you use a clean up tool like gtfstidy before reading the feed.

[leonardehrenfried]

I'm not a maintainer though. @sheldonabrown is.

[tomershefi]

@leonardehrenfried -
this is exactly why I need this change, to clean up & fix GTFSs
I am a developer at Moovit, we want to contribute to this project as well - where we process huge amount of GTFSs daily, using this library as well.

@sheldonabrown what do you think?

thanks