We should put the instruction_addr_adjustment into stack traces according to https://github.com/getsentry/rfcs/blob/main/text/0043-instruction-addr-adjustment.md

It looks like the libunwindstack that we use for unwinding on Android already does its adjustment before handing us the PC values:

https://github.com/getsentry/libunwindstack-ndk/blob/90d80b7073ae2108f7f42e2f576bddf988069ba2/Unwinder.cpp#L97

We should thus use "none", to avoid double-adjusting in the symbolication pipeline which could lead to wrong symbolication results.