Consider using a regex for grant id to prevent strange chars

[michaelwood]

Grant ids that contain strange chars have a way of breaking URIs and other analysis tools, it would be good to be able to prevent people from publishing the data. We can fix this at a tool level however this would mean that the identifiers aren't easily traceable from the original data.

ThreeSixtyGiving/grantnav#215
ThreeSixtyGiving/datastore#217
https://json-schema.org/understanding-json-schema/reference/regular_expressions

[michaelwood]

Note this would make the DQT additional check for these chars redundant

[mrshll1001]

We can definitely explore this as part of work on Version 2.0.

I really like this as a means to limit a charset for a grantid such as excluding spaces, backslashes, etc. as it means that the data will be more parseable by standard coreutils as well as more complex systems such as grantnav.

We could go as far as to mandate a particular pattern/prefix for grant ids, but I think the benefit of that would be more limited. I like the idea of starting light by making sure that the grant ids contain friendly characters.