refactor: deduplicate charset encode/decode into shared lookup map

[AnyCPU]

this is a little bit of refactoring to make it easier to maintain charsets related stuff.

pros & goals:

• single source of truth prevents charset bugs from requiring two-file fixes;
• eliminate duplication of 24 identical charset mappings across two files;
• reduce ~220 lines of repetitive switch/case boilerplate.

cons:

• negligible performance difference is dwarfed by encoder allocation and network I/O on every call.