externals: Update dynarmic to fc6b73bd

Resolves issues:
* 128-bit exclusive writes on Windows
* Non-updating CNTPCT_EL0

fc6b73 a64_emit_x64: Ensure host has updated ticks in EmitA64GetCNTPCT
888c67 a64_emit_x64: Fix stack misalignment on Windows for 128-bit exclusive writes
352d53 emit_x64_aes: Eliminate extraneous usage of a scratch register in EmitAESInverseMixColumns()
ab7fe7 A64: Implement SADDLV
09bd2b A64: Implement UADDLV
62e86d fp: Use forward declarations where applicable
b3edb7 emit_x64_vector: Append 'v' prefix onto movq in AVX path

file_util: Remove goto usages from Copy()

gl_shader_decompiler: Remove unused state tracking and minor cleanup.

gl_shader_decompiler: Implement SEL instruction.

file_util: Minor changes to ScanDirectoryTree() and ForeachDirectoryEntry()

Improvements to rasterizer cache

We can just leverage std::unique_ptr to automatically close these for us
in error cases instead of jumping to the end of the function to call
fclose on them.

This avoids a truncating cast on size. I doubt we'd ever traverse a
directory this large, however we also shouldn't truncate sizes away.

Avoids unnecessary copies when building up the FST entries.

file_util: Remove redundant duplicate return in GetPathWithoutTop()

video_core: Use nested namespaces where applicable

common: Remove synchronized_wrapper.h

file_util: Use an enum class for GetUserPath()

GPU: Implement the NVGPU_IOCTL_CHANNEL_KICKOFF_PB ioctl2 command.

This behaves quite similarly to the SubmitGPFIFO command. Referenced from Ryujinx.
Many thanks to @gdkchan for investigating this!

Instead of using an unsigned int as a parameter and expecting a user to
always pass in the correct values, we can just convert the enum into an
enum class and use that type as the parameter type instead, which makes
the interface more type safe.

We also get rid of the bookkeeping "NUM_" element in the enum by just
using an unordered map. This function is generally low-frequency in
terms of calls (and I'd hope so, considering otherwise would mean we're
slamming the disk with IO all the time) so I'd consider this acceptable
in this case.

Given both operands are the same type, there won't be an issue with
overload selection that requires making this explicit.

This is entirely unused in the codebase.

partition_filesystem, vfs_real: Minor changes

arm_interface: Remove unused tls_address member of ThreadContext

externals: Update dynarmic to 7ea1241

tests/arm_test_common: Minor changes

gl_shader_manager: Remove unimplemented function prototype

file_sys/errors: Remove redundant object constructor calls

gpu: Rename Get3DEngine() to Maxwell3D()

CPU: Save and restore the TPIDR_EL0 system register on every context switch

vfs: Minor changes

Resolves an issue with TPIDR setting being erroneously removed in the
dead code pass.

Loader: Only print the module names and addresses if they actually exist.

Given we're already constructing the error code, we don't need to call
the constructor inside of it.

We already return by value, so we don't explicitly need to make the
copy.