C#: Update Roslyn and other pinned depenencies.

[michaelnebel]

A couple of interesting things.

Changes to decoding.

It appears that the decoder compiler argument wasn't respected (there is an error that the code page is not installed when running the extractor on main) prior to this change. However, with the Roslyn update windows-1252 is added as an Emit option for the CSharp Compilation in the extractor.

If I decode the SBCS.cs in VS-Code using Windows-1252, then it is the character

Character: ’
Unicode: U+2019
Windows-1252 byte: 0x92

That is, the output of the test is now U+2019 (unless we change the content of the file).

Im am not sure exactly what the test is supposed to test. Is the test supposed to test that

• Decoding of 0x92 behaves differently under Windows-1252 than UTF-8? If that is the case, then expected output file has always contained a faulty value.
• Decoding a UTF-8 with BOM encoded file using Windows-1252 still uses UTF-8 even though Windows-1252 is specified as codepage? If that is the case, then the encoding of the file is wrong.

I assume that the test is suppose to test the latter. As such, I have updated the test file with the proper encoding (UTF-8 w. BOM).

Deriving unbound generic extensions methods.

It appears there has been a slight change in the symbols produced by Roslyn, which meant that we needed to do a small update to the logic handling extraction of extension methods.