Guidance on pattern to improve context for completions in an IDE

## Background

Our IDE has an opt-in feature where it will tell copilot about all files in the current project by doing `textDocument/didOpen` on them. This could potentially be dozens, possibly even hundreds of files, given the nature of the projects in real-world use.

This was originally implemented using the vim copilot plugin, now switched to using `copilot-language-server`.

## Question

Is this a "Good Idea"? Will it actually improve completion suggestions by providing more context, or will it quickly hit some internal limit?

When I switched this to `copilot-language-server` I also implemented setting `workspaceFolders` to true, and pass the root folder of the project as the workspace folder; is that accomplishing anything in terms of improving awareness of other files in the project (i.e. does this make the "pretend to open all files" approach redundant)?

Thanks!

Examinez le comportement du plugin vim copilot, le passage à copilot-language-server ainsi que la gestion de workspaceFolders et de textDocument/didOpen décrite dans l’issue. Déterminez si l’ouverture de chaque fichier du projet améliore le contexte de complétion ou atteint des limites, puis documentez le modèle recommandé et indiquez si workspaceFolders rend cette approche redondante.

Rédigé par le modèle d'indexation à partir du texte de l'issue.