Update cmp_tool to version v0.13

Added

• added chunk-specific compression parameter guessing functionality
• added chunk compression parameter file I/O functionality
• added fuzz target for decompression
• added Github action to compile with Werror and run tests
• added unit tests to improve code coverage

Changed

• improved spillover threshold calculation logic for chunk compression
• downgraded reserved field check from error to warning in decompression
• added const qualifiers for (de)compression parameters
• renamed CMP_ERROR_SMALL_BUF_ to CMP_ERROR_SMALL_BUFFER
• replaced dynamic max_used_bits struct with constant definition
• separated RDCU configuration into its own struct
• consolidated l_fx_variance and l_cob_variance into single l_fx_cob_variance field
• changed return type of compress_like_rdcu from int32_t to uint32_t
• updated Doxygen configuration and comments
• updated doxygen-awesome-css submodule to latest version
• fixed spelling errors and grammar in documentation and messages

Fixed

• corrected size calculation when dst is NULL in compress_chunk()
• fix be24_to_cpu function for big-endian systems
• fix collection size validation in decompression header

Removed

• remove deprecated ICU compression interface
• remove (de)compression from fast cadence data products