Make property/file content stringification type-safe

[orangeruan128]

Summary

Make .hex() calls on property/file payloads type-safe so a single corrupted property cannot abort parsing of an entire .one file.

Three call sites currently assume the payload is bytes/bytearray and call .hex() directly:

• pyOneNote/FileNode.py — FileDataStoreObject.__str__ (line ~559)
• pyOneNote/FileNode.py — PropertySet.get_properties() UTF-16 decode fallback (line ~671)
• pyOneNote/FileNode.py — PrtFourBytesOfLengthFollowedByData.__str__ (line ~762)
• pyOneNote/OneDocument.py — get_json() file-contents serialization (line ~73)

Symptom

On real-world .one files (large notebooks with many revisions, embedded files, or property variants), the payload at one of these sites can already be a str by the time stringification happens — typically because an upstream code path decoded a UTF-16 buffer in place, or a parser fallback returned a hex string. When that happens the call raises:

AttributeError: 'str' object has no attribute 'hex'

…which propagates out of process_onenote_file and stops parsing entirely, even though only a single property is malformed.

Fix

Keep .hex() as the path for bytes/bytearray (no behavior change for the common case). Add explicit branches for the two non-bytes cases that have been observed in the wild:

• str → pass through (or encode to bytes first in the OneDocument.get_json() case so the JSON output stays hex-only)
• anything else → repr(...) so the parser produces a readable placeholder instead of crashing

Scope / non-goals

• No new functionality.
• No changes to parser semantics for well-formed input.
• Does not touch the issues already covered by Implements candidate code for pocessing ArrayOfPropertyValues #16 (ArrayOfPropertyValues) or Ensure FileNode.data is initialized in __init__ and is not used if it… #20 (FileNode.data initialization). Those are separate, complementary fixes; this PR is intentionally narrow so it can be reviewed and merged on its own.

Verification

• ast.parse on both modified files passes.
• After applying this change on top of main, parsing of a corpus of OneNote files that previously failed on 'str' object has no attribute 'hex' completes successfully and produces the expected JSON / extracted file output.
• Files that already parsed cleanly continue to produce byte-identical JSON for the headers/properties/files structure.