chore: use initLog() in utils, to handle use case when LSP server is not loaded

[wiedld]

Follow up to influxdata/flux-lsp#618

The LSP wasm was panicking due to use of a logger which was not compatible with wasm. The recent LSP version bump included:

1. a global wasm logger init whenever the LSP server was init.
   • no UI change required, outside of the version bump
2. the option to manually init the wasm logger, when the LSP utils methods are used without the LSP server being init first
   • this does require calling initLog() in the UI, to handle race condition if LSP server is not loaded --> but LSP utils methods are used
   • setting the global logger >1 time, has no negative impact. (Only resulted in console warning within the LSP tests.)

Impact in UI.

We do have a current, frequent bug in the UI where the dashboard fails after a panic occurs in the LSP wasm. As we did not know the root of the panic (and we didn't have any error logs from the wasm), we focused instead on the cascading nature of the variables tie-in with dashboards.

Recently, we became aware of that fact that our lack of logging from wasm in prod (especially during instances of wasm panic) was directly due to our logging causing the panics. This is a followup PR to ensure that the global logger is always set.

Checklist

• A PR description, regardless of the triviality of this change, that communicates the value of this PR
• Well-formatted conventional commit messages that provide context into the change
• Documentation updated or issue created (provide link to issue/PR)
• Signed CLA (if not already signed)
• Feature flagged, if applicable

[hoorayimhelping]

is it possible to call this elsewhere? Side effects in modules like this should be avoided if possible. This function will be called anytime code imports parse or format_from_js_file. Is it possible to move this function call into the parse and format_from_js_file functions?

[wiedld]

Since the LSP doesn't have a single entry point, @hoorayimhelping and I agreed to leave this code as the best alternative. Doc'ing the tech debt in the LSP, that we need some refactoring to enable the single entry point.