Modified Secrets Handling to have Alternative Windows Codepath

[gosselind1]

Based on discussion in #695 I have gone ahead and implemented an alternative secrets object for windows, using pywin32.

To summarize the changes:

1. Secrets providers have been removed from util.py, and move to their own Secrets.py module.
2. Secret service object is now provided by the module rather than the util module, and imports have been updated accordingly.
3. Added a function to determine if the application is operating under a msys2 environment.

I've done some testing on windows, but I haven't gotten around to testing this on linux. These changes shouldn't have any impact though.

There also might be some oddities if the system's group policy has been modified to a non-standard configuration, though I've attempted to handle those edge cases.

[TingPing]

I don't believe this does any blocking IO or anything, you can just simplify it down to:

def is_msys2() -> bool:
    return sysconfig.get_platform().startswith("mingw")

The caching on Linux is because it access the filesystem (but honestly probably doesn't matter in Python much anyway).

[TingPing]

Overall this looks OK. I need to find time to test it on Linux myself.

I do question why its limited to msys2 and not all of Windows?

[gosselind1]

I'm using msys2 to run the program on windows, but could be possible without it. This would depend on dependencies and other things. Currently I'm using msys2 to resolve most of the library dependencies, and other software components though.

To check broader more native windows compatibility, I would need to manually compile and install a lot of different pieces of software, but it might be possible? I could change the current utility function to a more generic windows one for now whilst investigating and testing this.

My initial motivation with using msys2 was due to what appeared to be a medium percent of linux related software components that might not have pure windows implementations and or compatibility.

[TingPing]

The whole stack supports being built with msvc. You don't really need to test it, I just mean change the check to is_windows() entirely.

[gosselind1]

Ah. Let me clarify, I had wrote it to target msys2 because I was unsure if it had full native windows support, as I've been getting it to build via msys2. I'll change the function to generically target windows.

[gosselind1]

I got a moment to poke through this again. What's the specific build process for flatpak? I wanted to replicate the failing build, but the flatpak configuration is crashing on code unrelated to the changes I made, and I do not think this is the issue the preference dialog is failing on either.