feat: Reduce severity of SameNameAndDescriptionForRouteNotice to WARNING

[aababilov]

A route that has the same route_desc and route_short_name or
route_long_name may be still served to the end user, so this is not an
error. In fact, many consumers may just ignore route_desc completely.

Numerous feeds that are consumed by Google do have this notice and they
are considered valid.

We also update the docs for the notice.

[isabelle-dr]

The specification is currently being reviewed to become RFC 2119 compliant (see changes here).
The ambiguity in routes.route_desc is planned on being solved by having:

Must not be a duplicate of route_short_name or route_long_name.

If this does not make a dataset invalid (and should trigger a WARNING in the validator), we should:

1. Modify the change planned to the spec in PR227 to treat as a warning
2. Modify the validator severity from error to warning to match the spec

@MobilityData/transit-specs thoughts?

[aababilov]

I have posted a comment regarding route_desc on google/transit#277

I suggest to merge this pull request #917 first because the current behaviour of Validator treats plenty of feeds as invalid. This makes more difficult adoption of the Validator by members of GTFS community.

[isabelle-dr]

Hello,

Thanks for your comment on google/transit!
I'm okay with merging this PR first @lionel-nj

[lionel-nj]

Thanks for working on that @aababilov! One small documentation change should be provided here: SameNameAndDescriptionForRouteNotice should be in the table of warnings instead of the table or errors in RULES.md (same thing for the descriptions) and /docs/ NOTICES.md :)

[aababilov]

Thanks for working on that @aababilov! One small documentation change should be provided here: SameNameAndDescriptionForRouteNotice should be in the table of warnings instead of the table or errors in RULES.md (same thing for the descriptions) and /docs/ NOTICES.md :)

Done. I think that would be great to develop a way for generating that doc from Java code.

[lionel-nj]

Thanks @aababilov!

I think that would be great to develop a way for generating that doc from Java code.

I agree, that would certainly help! Do you have examples of projects that have implemented this process in the past? Will definitely talk about this with @isabelle-dr and see how we can improve the gtfs-validator to this extent. Thanks for the suggestion!