feat: Capture registers with inproc backend (FEEDBACK-1413)

[Swatinem]

No description provided.

[supervacuus]

I tested this on my m1 mbp and added register writes via inline assembly right before the crash and could see most of the general-purpose registers survive into the disk dump of the event. I think you tested it on all OS+Arch combos I could provide besides that. But please let me know if further target-testing makes sense.

[Swatinem]

I tested this only on x64 on all OSs. I would like to have someone give it a spin on 32/64-bit Android as well.

[supervacuus]

I can test Android only on the emulator and give aarch64 on Linux a try. But both probably not before Wednesday.

[supervacuus]

I tested this on aarch64 under Linux where it works correctly. I also tried to test this in android using the emulator. I can see that an aarch64 register dump is created in the sentry-native database path on the device, but I cannot really stress the register values for comparison because the crash-causing function is compiled using an armeabi-v7a toolchain. My insight into the android build setup is still lacking, but I will investigate further.

[supervacuus]

android aarch64 seems to work too. I cannot test armeabi-v7a, because the images provided for the emulator are too old. I do not own a device.

[Swatinem]

@supervacuus Thanks for the investigation! I think this is ready to land in that case.