This adds a small module that will annotate a list of targets that we're
about to scan with relevant info, and now we will send that off in
traces.

Note that most of this PR is just adding yojson derivers to things. If
we don't like the way I chose to derive some of the edge cases I'm open
to changing them!

## Test plan
```bash
semgrep --pro --trace --config p/default --trace-endpoint semgrep-dev
```
in https://github.com/SigNoz/signoz. relevant trace
[here](https://jaeger-dev2.corp.semgrep.dev/trace/53dd39781d191adb57711cc43f518608?uiFind=b704e985bd5b28c9),
check the `annotated_targets` tag, and hit "copy" not "JSON".

Easiest way to use it is something like pbpaste:
```bash
pbpaste > annotated_targets.json
jq ".[] | select(.stat.textual | not) | .internal_path" annotated_targets.json # get list of non text files
jq "sort_by(.stat.line_count) | .[] | select(.stat.line_count > 4000) | {path: .internal_path, line_count:.stat.line_count}" annotated_targets.json # get list of files w/ line count > 4k
jq ".[] | select(.minified) | .internal_path" annotated_targets.json # get list of minified files
```

tested on elastic search also (>30k targets). Only adds 5s overhead
there. Jaeger is still workable too, and doesn't lag unless manually
exploring the json. Note we will need to wait until @semgrep/infra bumps
some jaeger settings though before traces with >5k targets will actually
be picked up since they're relatively large.

synced from Pro b942b9a1266d7975333a74f03ad1ae3c24def812