Workspace trust prompt improvements

[TheGLander]

Introduces some improvements to how workspace trust prompt is handled:

• Added editor.insecure-lsp config option, which allows LSPs to start, without loading local configs
• Disabled showing the trust prompt for when there's nothing we need trust for. This means that if you have editor.insecure-lsp enabled, you will only ever see the prompt when there are local configs
• Documented editor.insecure and editor.insecure-lsp in config reference and the trust prompt

These changes are unrelated to #15582

[the-mikedavis]

This looks like a nice improvement, both not showing the prompt unless it's required and being able to block config but implicitly trust language servers 👍

I just have an opinion on the config

[Rudxain]

This means that if you have editor.insecure-lsp enabled, you will only ever see the prompt when there are local configs

That could be extended so that if insecure-lsp is disabled, Helix only prompts when it's going to auto-start a LS. This is useful because some people only have a handful of LSs installed.

That might be better for a follow-up PR, to keep this scoped.

unrelated to #15582

The title of my PR is very ambiguous 😅. I'll change it

[TheGLander]

Merged editor.insecure and editor.insecure-lsp into one string option, as requested. Also fixed an oversight where the prompt wouldn't be shown if language servers are disabled and there're local configs, due to me not removing the now-obsolete language server check.

Also, yeah, I think it would make sense for conditional LS trust check to be a separate PR because it looks like it'd require some changes to how language server are loaded?

[David-Else]

It makes no sense for the config variable to be called insecure anymore now it can hold the value of none, lsp, or all. I think it should be workspace-trust-level. You can then remove the redundant The level of trust every workspace is given. from the docs.

[TheGLander]

Renamed the option to editor.workspace-trust-level as requested