Add UTF-8 unit tests for apteryx_set, apteryx_get, apteryx_set_tree, apteryx_get_tree

[Copilot]

Adds a dedicated "UTF-8" CUnit suite (19 tests, 132 assertions) covering both valid multi-byte Unicode and known edge-case / invalid-input behaviour across all four core data APIs.

Happy-path tests (10)

Verify correct round-trip storage and retrieval — including byte-length assertions to catch silent character-to-byte collapsing:

Coverage	APIs
2-byte Latin Extended (é, ö)	set/get
3-byte CJK (日本語)	set/get
4-byte emoji (😀)	set/get
2-byte Arabic + Cyrillic	set/get
Mixed widths (café 日本 😀, 17 bytes / 9 chars)	set/get
Multi-byte char in path component	set/get
UTF-8 leaf values via tree	set_tree/get
UTF-8 values returned in tree nodes	set_string/get_tree
UTF-8 node keys	set_tree/get and set/get_tree

Edge-case / invalid-input tests (9)

Document two known limitations of the current implementation:

Class A – Silent NUL-byte truncation.
rpc_msg_encode_string() uses strlen(), so any embedded 0x00 terminates the data at that point; bytes after it are silently dropped.

char value[] = { 'c','a','f', 0xC3,0xA9, 0x00, 'm','o','n','d','e', 0x00 };
apteryx_set(path, value);
result = apteryx_get(path);
// strlen(result) == 5  — only "café" survived; "monde" was lost

Covers: value truncation (set/get), path truncation (write redirected to wrong key), and tree-leaf value truncation (set_tree/get).

Class B – No UTF-8 validation.
validate_path() and the RPC layer perform no byte-content inspection. Invalid sequences are accepted and returned unchanged:

Sequence	Why invalid
\x80	Lone continuation byte
\xC1\x81	Overlong encoding of U+0041
\xE2\x80	Truncated 3-byte sequence (EM DASH missing 3rd byte)
\xFF	Never valid in UTF-8
\xED\xA0\x80	UTF-8 encoding of surrogate U+D800
\xC3\xA9\x80	Valid 2-byte prefix + lone continuation byte (via set_tree/get_tree)