add a flag to wrap models/tests in jinja blocks

[beckjake]

And then run all tests through that. Everything seems to have worked fine!

Fixes #1400

[cmcarthur]

this is interesting, but it seems almost tautological.... unless someone included an unclosed comment or a {% endmodel %} in their model i would expect this to always pass. I am more interested in just running the toplevel block parser over the unwrapped model contents, and comparing any compiler errors to those raised by the jinja parser. if the block parser raises an error, but the jinja parser does not, then we have a bug that we need to fix.

you should do the exact same thing for macro files too. --wrap-models-in-tags is probably a bad name if this turns the parser on for macros and models, so maybe --test-new-parser or something like that would be a better name

[beckjake]

unless someone included an unclosed comment or a {% endmodel %} in their model i would expect this to always pass

Well, me too, but I thought the idea was to shake out bugs around top-level block extraction that are in that category. I guess I misunderstood - I'll change it to be the way you've described, which seems like a reasonable enough test.

[cmcarthur]

@beckjake gotcha. I care more about comparing our parser to jinja's parser on arbitrary jinja code, not necessarily testing model block extraction. unrelated: i think wrapping model files in model blocks is a good way to achieve forwards compatibility with this new parser.

[beckjake]

Yup! I that is the longer-term plan, I think. Maybe we can get it in for wilt-chamberlain if there isn't too much already, that would be nice.