PyCForge 1.3.11 — Normal startup regression correction

Correction
An explicit close-and-reopen check found an incorrect Startup report in 1.3.10 after saving an empty workspace. Qt writes an empty list as @Invalid() in INI settings. A fresh process reads it as None; the in-process settings cache can retain the original list type. The open-file and recent-file readers incorrectly classified this valid empty representation as damaged settings.

The readers now accept the valid empty representation. Normal startup creates no Startup report when the settings and previous files are usable. Genuine invalid settings and unavailable files still produce the recovery report. Explicit --safe-mode continues to display its intentional safe-start notice.

Verification
- Six complete-launch normal-startup cases passed against the source tree and again against the installed 1.3.11 wheel: fresh launch, restart of an empty workspace, upgrade using genuine 1.3.9 settings with valid files, restart with saved geometry and preferences, and valid settings schemas 1 and 2.
- Each normal-startup probe ran the actual Qt startup checks and event loop in a separate process. It watched for report-window show events and also verified no report was created, no settings issues or restoration failures were recorded, the expected files and active file were restored, and no QTextCursor position warning appeared. No unexpected report appeared in any of the six cases.
- Source files used in the probes remained byte-for-byte unchanged. The valid files included Unicode names and a directory with spaces.
- Automated normal-startup, corruption recovery, previous-file restoration, startup progress, and related Qt regression checks: 53 passed, 18 subtests passed. Six dedicated no-report tests and their subprocess probe are included in tests/.
- Installed-wheel corruption checks confirm that real damaged settings still produce the white, read-only report; clipboard copying works; damaged settings bytes are preserved; and safe start avoids accessing the original store.
- Installed wheel: 825 application files match source, 834 RECORD hashes verified, 495 icons retained, and all 21 Help pages opened. Typed conversion and a native GCC single-inheritance comparison passed.
- Offline conversion corpus: 37 cases, 29 tested matches, 8 expected unsupported cases, no unexpected results or observed network attempts. All 335 recorded application source hashes match this release.
- README and HTML Help content are unchanged from 1.3.10. The existing Help describes the intended normal and recovery behavior.

Limits
Checks used Linux, Python 3.12, Qt offscreen, and GCC 13.3. Native Windows verification remains outstanding. This is evidence for the tested cases, not a claim that all application bugs have been eliminated. The full historical suite was not rerun, and nothing was published to PyPI.
