Fix long browser filenames

[chrisgleissner]

Fixes #715.

Overview

This PR fixes long filename handling in the Ultimate 64 browser when acting on FAT-backed entries.

User-visible failures before this change:

• Mount Disk and Run Disk could fail with Opening disk file failed.
• Rename could fail with ERROR: FILE DOESN'T EXIST.
• The browser could display long filenames correctly, but still fail to act on them.

Bugs

1. BrowsableDirEntry::getName() converted browser action names through a fixed 64-byte buffer.
2. FileInfo::generate_fat_name() did not guarantee NUL termination for direct FAT names.
3. The shared filesystem path walker still matched entries through a fixed 64-byte scratch buffer, so long direct names could not be resolved reliably even after the browser handed them off.

Fixes

• Return direct FAT names from BrowsableDirEntry::getName() without truncating them.
• Keep CBM-name conversion behavior unchanged, but size that conversion buffer from the actual entry length.
• Make FileInfo::generate_fat_name() always NUL-terminate direct-name output.
• Widen the common FileSystem::walk_path() FAT-name scratch buffer and pass its full size into generate_fat_name() so shared path resolution handles long names end to end.

Tests

Host unit / regression tests

• make -C software/filemanager/tests temp-auto-cleanup
• Added GenerateFatNameKeepsLongDirectNames
• Added FileManagerHandlesLongDirectNames

Real hardware / e2e verification

The following e2e test was performed against an Ultimate 64 Elite I.

It successfully demonstrates rename and mounting of filenames exceeding 64 chars, specifically the name zzzz_long_filename_browser_regression_0123456789_0123456789_0123456789_0123456789.d64.

Invocation: ./tools/io/filemanager/browser_long_filename_test.py

Output:

[01] seed long-name fixture for rename ... OK
[02] browser rename on long filename ... 
[03] open menu ... OK
OK
[04] reseed long-name fixture for mount ... OK
[05] browser mount on long filename ... 
[06] open menu ... OK
OK
browser_long_filename_test: OK

[chrisgleissner]

Hi @GideonZ ,

as mentioned on #715 , this PR fixes the issue as shared by its reporter.

Thanks for review and merge.

Best,
Christian