Description
When building on RISC-V, allow users to opt in to unaligned access by setting the UNALIGNED_ACCESS_SUPPORTED This appends
-DUNALIGNED_ACCESS_SUPPORTED to CFLAGS, which in turn enables LZ4_FORCE_MEMORY_ACCESS=2 under RISC-V only.

Usage:
make UNALIGNED_ACCESS_SUPPORTED=1

Reason for modification
During testing, we found that direct access to unaligned memory is currently the best method on RISC-V, similar to the method on ARMv6. However, not all RISC-V servers support unaligned memory, and the GCC compiler does not have a macro to distinguish whether unaligned memory is supported. Therefore, we chose to add a new macro to manually allow users to enable direct access on RISC-V servers that support unaligned memory to achieve performance improvement.

Testing
Testing of this modification on a RISC-V server found that the compression rate increased by 30%. The test data is as follows:
Tool: lzbench
Unaligned_Access_Supported disabled:

lzbench 2.1 | GCC 13.2.1 | 64-bit Linux |

Compressor name         Compress. Decompress. Compr. size  Ratio Filename
lz4 1.10.0               96.8 MB/s   206 MB/s   100881340  47.60 ../silesia.tar
[Params] cIters=5 dIters=5 cTime=0.0 dTime=0.0 chunkSize=1706MB cSpeed=0MB

Unaligned_Access_Supported enabled:

lzbench 2.1 | GCC 13.2.1 | 64-bit Linux |

Compressor name         Compress. Decompress. Compr. size  Ratio Filename
lz4 1.10.0                124 MB/s   209 MB/s   100881340  47.60 ../silesia.tar
[Params] cIters=5 dIters=5 cTime=0.0 dTime=0.0 chunkSize=1706MB cSpeed=0MB