refactor: new workspace crate

[psteinroe]

current status / next todo:

• understand and decide on settings / configuration / workspace setup
• make lsp and cli with server work: load config, open and close files
• migrate change from base_db to workspace_new
• make change file work e2e
• make db connection and schema cache work.

copying (with acknowledgements) from biome to get things going

• pg_markup
• pg_console
• pg_text_edit
• pg_diagnostics_macros
• pg_diagnostics_categories
• pg_diagnostics
• pg_configuration

will serve as a base to work on

• pg_workspace_new (to be renamed to pg_workspace later)
• pg_cli

goal of this pr is to be able to start and stop the daemon process and accept file changes, connect to the database and populate the schema cache. we can then start moving features over in follow-up prs.

we will use the same client-server architecture as biome, but with a few adaptions to make it work for postgres. the list is not complete, just thoughts gathered while working on this:

• biome spawns a single daemon per machine / workspace. each project registers itself. this works great for them because all configuration and all analysis results only depends on the file system. for us its a bit different, because a type checker will report different result for the same file if its pointed at different databases. hence, we will spawn one daemon per target database, and there is no project registry - just a single one. this will also work well when being integrated in platforms such as supabase: there could be a single lsp server running for each project.
• I will use the toml format for the config file. it will contain only the db connection string for now, but the foundation for all sorts of configurations is laid out already.