Change workaround for the dynamic core

[henk717]

The existing workaround for the dynamic core hangs the process instead of crashing, it made the workaround worse than the original situation. This PR undoes the problematic commit and replaces it will a full restart command automatically relaunching the process.

What issue(s) does this PR address?

#5584

Does this PR introduce new feature(s)?

No

Does this PR introduce any breaking change(s)?

The entire software is now restarted, this does not break for me but I can't rule this out on more exotic configurations.

Additional information

I totally get it if you'd rather not merge this quick workaround, but as this Daum style restart solves the issue for me I wanted to offer the changes to motivate the issue being fixed.

[joncampbell123]

Odd, I've been using the dynamic core code in question every time Windows 98 wants to reboot the system without any problems, Linux x86_64.

[joncampbell123]

The idea with how DOSBox-X does it is that it signals the desire to reset and then lets the CPU core finish executing instructions.

Throwing a C++ exception with dynamic core may happen to work in Win32 but for most platforms, the C++ runtime cannot unwind the stack properly because it doesn't recognize anything that was dynamically generated.

If that method is causing DOSBox-X to hang, then additional work needs to be done so that on the CPU reset signal the core or normal loop needs to make sure it exits the core loop and then throws the desired exception.