xtensa: Fix backtrace on exceptions

[Ouss4]

Summary

Full backtrace was not available, because the code responsable for retrieving the correct registers (A0 and A1) was reading them from an incorrect SP.

Impact

Xtensa

Testing

before:

(gdb) bt
#0  xtensa_assert () at common/xtensa_assert.c:129
#1  0x400d402e in xtensa_user_panic (exccause=29, regs=0x3ffe0b88) at common/xtensa_assert.c:311
#2  0x400d39a6 in xtensa_user (exccause=29, regs=0x3ffe0b88) at chip/esp32_user.c:438
#3  0x4008086f in _xtensa_user_handler () at common/xtensa_user_handler.S:276
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

after

(gdb) bt
#0  xtensa_assert () at common/xtensa_assert.c:129
#1  0x400d402e in xtensa_user_panic (exccause=29, regs=0x3ffe0b88) at common/xtensa_assert.c:311
#2  0x400d39a6 in xtensa_user (exccause=29, regs=0x3ffe0b88) at chip/esp32_user.c:438
#3  0x40080875 in _xtensa_user_handler () at common/xtensa_user_handler.S:282
#4  0x400dd66a in hello_main (argc=0, argv=0x0) at hello_main.c:40
#5  0x400d2607 in nxtask_startup (entrypt=0x400dd65c <hello_main>, argc=1, argv=0x3ffe0480) at sched/task_startup.c:70
#6  0x400d13e3 in nxtask_start () at task/task_start.c:133

[Ouss4]

@gustavonihei I added one small commit since last review, PTAL.