fix: Attempt to send minidumps for all process-gone reasons

[timfish]

It was reported that webContents.forcefullyCrashRenderer() results in a minidump being generated but render-process-gone notifies that the process was killed.

We were ignoring all reasons which were not crashes and this resulted in ignoring some minidumps and then sending them on next app start and assuming they came from the main process.

This PR:

• Looks for minidumps for all exit reasons
• Includes the exit.reason in tags so customers can filter them

[Cefold]

After webContents.forcefullyCrashRenderer() event, there was a minidump under pending folder, but it seems that Sentry cannot find it. It will upload it until next app restart, thus event.process will be classified as browser instead of renderer.

Sentry Logger [log]: 'renderer' process 'killed'
Sentry Logger [log]: Found 0 minidumps

[timfish]

but it seems that Sentry cannot find it

It looks like the issue is that the minidump does not exist when render-process-gone is called. I need to do some more testing!