Allow partial reads and align with GTFS specifications for stop_lat, stop_long, and transfer_type

[InterferencePattern]

I'm not sure if you're interested in this change of behavior, but it may be valuable to read a GTFS file partially instead of failing due to a single malformed record.

Additionally, changed validators to match reference

• Allows 4 and 5 as transfer_type
• Allows None for stop_lat and stop_long

[jarondl]

Thank you so much for the contribution!

[jarondl]

Doesn't this make all of the in_range validated fields nullable? What am I missing?

[InterferencePattern]

Since some floats are nullable, this rule was previously too strict, but you're right- this solution may be overly broad.

I can look into adding a nullable argument so this check won't fail only for nullable columns.

[InterferencePattern]

Please check this, I think I've made the necessary change to at least allow a check on nullable columns. I still need to make it conditional on the value of another column (but I'm new to sqlalchemy)

[InterferencePattern]

Do you have any examples where you've successfully performed conditional checks? The checks I see only require one column at a time.

For stop_lat and stop_long:

Conditionally Required:

• Required for locations which are stops (location_type=0), stations (location_type=1) or entrances/exits (location_type=2).
• Optional for locations which are generic nodes (location_type=3) or boarding areas (location_type=4).

[jarondl]

No I don't have examples for that. I guess we can simply be lenient here.. I'm not a fan of standards with optional required fields.