Separate functions for 32-bit linear memory

[daxpedda]

Currently decodeStringFromUTF8Memory() and encodeStringIntoUTF8Memory() take and return i64s to support 64-bit linear memories. This requires integer conversions when working 32-bit linear memories, which has minimal overhead relative to string de/encoding.

The issue has been brought up in a CG meeting and it was proposed to:

• Add overloads with i32s.
• Add separate functions with separate names with i32s.
• Do nothing.