Add inline syntax highlighting for awk with `zed-awk`

[rosingrind]

Redirected there from dangh/zed-awk#1. Suppose you have a part of this sample shell script:

$_comp_command1 2>/dev/null |
    sed -r 's/(-[[:alpha:]-]+, )+(--[[:alpha:]-]+)+/\2/' |
    sed -r 's/^[[:space:]]+([[:alpha:]-]+)+[[:space:]]+/\1|/' |
    grep -E '^(--)?[[:alnum:]-]+\|[[:alpha:]]' |
    awk -F '|' '{ $2 = toupper(substr($2,1,1)) substr($2,2); print $1, "|-- ", $2 }' |
    column -s '|' -t

I'm talking about highlighting this specific part:

{ $2 = toupper(substr($2,1,1)) substr($2,2); print $1, "|-- ", $2 }

[notpeter]

This might be possible, but I believe it would require tree-sitter-bash to create sed/awk/grep specific captures so they could be used as injection points and then This is for example how Zed is able to render language specific injections for markdown.

```javascript
var a = 123
```

Practically speaking, it's a little more difficult because it's not just an explicit ```(lang)\n([.+])``` capture blob, it has to capture the command, parse any command line arguments/variable expansions/quoting/etc and then annotate the single quoted string later with with that specific command for the command-specific injection be defined. My guess is that it's possible, but I'm not a tree-sitter expert so I wouldn't know exactly where to start on the tree-sitter side.

Some examples to get you started:

• javascript injection for regex crates/languages/src/javascript/injections.scm
• markdown dynamic injection based on language following triple backticks: ```lang
  • crates/languages/src/markdown/injections.scm
  • tree-sitter-markdown/grammar.js