Escaping fix + test suite

[jonascarpay]

Fixes #35
Fixes #5

This PR contains a rudimentary test suite, as well as a fix for the escaping issue.

• More tests for string edge cases
• Don't actually just show in Haskell the more I think about it the more I think this is fine
• Run on CI
• Better type for run

[jonascarpay]

I'm on mobile, which doesn't show when or how this got marked as ready for review, and I can't change it back. But just to be sure, it's not ready yet, I'll request review when it is.

[cdepillabout]

@jonascarpay No problem, I switched it back to draft for you.

[cdepillabout]

I don't think it really matters for this PR, but I thought you might be interested that it seems like PureScript people like to use empty data types for things like Void, but use foreign import data for things that are coming from FFI:

https://discourse.purescript.org/t/what-does-using-data-without-any-assignments-mean-i-e-data-something/1651/5?u=cdepillabout

[jonascarpay]

After some offline discussion, we decided that the test suite needs a better design before being worth merging.

The issue with this test suite is that, while it looks like a suite of unit tests, it acts more like an integration test. Put differently, the most interesting part about it isn't that it passes all the tests, but that it runs at all. Either we acknowledge that and have a design that better communicates that fact, or we need to find a way of better isolating the guarantees that we want our test suite to provide.

[m-bock]

hi @jonascarpay
cool, that you are adressing those issues! Today I got aware of a thing, that may be helpful:
On main the commit 72cc3ff introduces a bug that causes escape sequences like \x001b[32m to be printed as ESC[32m when logged/traced to the commandline. It should actually turn the following text's color to be green instead.
I tried it with the f29c014 from your PR branch and it's the same behavior.
I cannot investigate details right now, If it's not related I can open up a separated issue.

[jonascarpay]

@thought2 Oh wow, that's a very good catch! Please open a separate issue, and I'll have a think how to best go about fixing it. I guess that with our strings passing through Nix, PureScript, Haskell (and arguably Bash) just using show, and that handling all our escaping concerns was too good to be true 🥲 .