[SAOC 2024] - Leverage dmd as a library in D-Scanner in order to extend its functionality - Weekly Update 6

[Vladiwostok]

Greetings,

I managed to fix the linting issues in phobos:
https://github.com/dlang/phobos/commit/354781f22985dbc6eb9bbe0584540047f85bf725
https://github.com/dlang/phobos/commit/231ae8b68ab5d4d81df4ffd4ec08c896b6c9ca81

I fixed another issue that popped up which made the testing 
pipeline fail. Some code from `errors.d` was moved in another 
file and that file has not been added in the dub recipe for the 
lexer subpackage.
https://github.com/dlang/dmd/commit/4a90885ef8e0fc29334da91a4318f9ab52af9f80
https://github.com/Dlang-UPB/D-scanner/pull/164
In D-Scanner we use commit hashes to specify which package 
version should be selected. However, dub only takes into account 
tagged versions, and since the commits were not tagged, the 
master branch was fetched instead. For now, we cannot pin the 
latest DMD version as we already implemented those changes. I'm 
thinking we should pin the DMD version on the next release.

I did some more cleanup:
https://github.com/Dlang-UPB/D-scanner/pull/166
https://github.com/Dlang-UPB/D-scanner/pull/163
https://github.com/Dlang-UPB/D-scanner/pull/162

I took a deeper dive into the Windows issue that was affecting 
D-Scanner for the past weeks. I think we are dealing with a bug 
in the DMD Lexer. The locations for tokens are calculated 
wrongfully on Windows (file offsets and line numbers). It has 
nothing to do with delimited strings. The issue was detected now 
because Autofix requires precise file offsets in order to insert 
replacement code.