Update hypothesis.R to allow for backquoted parameters

[stemangiola]

I happen to have this parameter ethnicity_groupsNativeAmerican&PacificIslander, for a series of fits (30K genes, with 5K samples each, in a biological study) that take ~ 2 weeks, to run.

The character & would normally cause an error.

Instead of rerunning everything, I have enabled hypothesis() to allow for backquotes, such as

hypothesis(
          c(
            "Europeans" = "(ethnicity_groupsAfrican
    + ethnicity_groupsEastAsian
    + ethnicity_groupsHispanicDLatinAmerican
    + ethnicity_groupsSouthAsian
    + `ethnicity_groupsNativeAmerican&PacificIslander`) / 5 = 0"

If this is side-effect-free, I think it would be a great addition to give more flexibility to users. Afterall the modelling worked with atypical names, it would make sense the whole pipeline would.

If you are interested I could design a unit test.

Note: This PR is based on a ChatGPT suggestion that I cleaned up.

[paul-buerkner]

I think this would be good feature. Thank you!

Adding some unit tests for this would be good. I suggest to add some in local tests (in tests/local/tests.models-5.R) would be great.

Can you also slightly edit the style to fit with the general code style of brms. That is, use <- instead of = for assignments. Use function(...) instead of function (...). Space before {.