Add language support for persistent library syntax

[lykahb]

This adds support for the language used by a persistent library to define the database models. This syntax is only used by one library, so I chose to prefix the language name with "haskell-" for specificity.

The haskell injections query uses the name of the function that handles that content of the quasi-quote as the name of the language. The persistent library has three of them that I included into the injection-regex.

[lykahb]

I also tried an alternative recognition of the injected language that moves the recognition from injection-regex into the haskell injection query. Let me know which variant you consider cleaner.

; haskell/injections.scm

((comment) @injection.content
 (#set! injection.language "comment"))

(quasiquote
 (quoter) @_quoter
 ((quasiquote_body) @injection.content
  (#match? @_quoter "(persistWith|persistLowerCase|persistUpperCase)")
  (#set! injection.language "haskell-persistent" )
 )
)

(quasiquote
 (quoter) @injection.language
 (quasiquote_body) @injection.content)

# languages.toml

[[language]]
name = "haskell-persistent"
scope = "source.persistentmodels"
injection-regex = "haskell-persistent"
file-types = ["persistentmodels"]
roots = []
comment-token = "--"
indent = { tab-width = 2, unit = "  " }

[lykahb]

Moved recognition of language name into the haskell injection queries, added folds, and marked = as an operator. Now the language support is closer to the one that got merged into nvim-treesitter.