If a directory contains a `.oct-config` file that specifies an encoding that is different from the default .m file encoding, the editor still tries to open files in that directory with the default encoding. That might lead to garbled characters in the displayed file.

See also: https://github.com/catch22/octave-doctest/pull/256#issuecomment-1183671829

I'm attaching a work-in-progress patch for that.
With it, files open in the encoding specified with `dir_encoding` or in the `.oct-config` file if it is already in the load path.

The `.oct-config` parsing logic is coupled to the load path (to take advantage of caching). That means that files from directories that aren't in the load path are opened with the default encoding even if their directory contains a `.oct-config` file.
Imho, that might be an ok trade off for performance.

Another thing that I noticed: If I leave one of those files with differing encoding open in the editor and close Octave, they are displayed with the default encoding (i.e., with garbled characters) the next time I open Octave.

@ttl: Do you know if it is possible to store the encoding for open files in the editor on a per tab basis?