Add install history recording and batch reinstall from history

[Copilot]

• Add record_install_history function to common.sh that appends install commands to $HOME/.config/pengwin/install-history.txt
• Call record_install_history in the main pengwin-setup script after successful install_menu completion (skip uninstall commands)
• Create pengwin-setup.d/batch-install.sh - batch install from history script with checkbox dialog
• Add "BATCHINSTALL" menu entry to pengwin-setup.d/maintenance.sh
• Update completions/pengwin-setup with MAINTENANCE completions
• Update help text with MAINTENANCE actions
• Create test file tests/batch-install.sh
• Add test to tests/run_tests.sh
• Run shellcheck on modified scripts
• Run code review and address feedback
• Fix failing test: use grep -cxF with specific patterns piped from run cat, matching existing test patterns in the suite

Original prompt

after each sucessful installation, add to a file, called for example $HOME/.config/pengwin/install-history.txt, the commands of the installation. For example for pengwin-setup install GUI TERMINALS TERMINATOR write to the file GUI TERMINALS TERMINATOR

Then add a functions in the MAINTENANCE menu called batch install from history or something like that. This function will present a checkbox list with all options (unique, if one is already don't add it again) and all the options will be checked by default. Then when the user press oK, pengwin-setup will be called in this order:

pengwin-setup update
pengwin-setup install [first line of checked options]
....
pengwin-setup install [last line of checked options]
message telling which options were installed and which failed. so if one fail you'll keep installing

at the end after the message closes, call wsl --terminate "${WSL_DISTRO_NAME}"

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[crramirez]

@copilot tests are failing

test_install_history_recorded
History content: SETTINGS MOTD MOTD_NEVER
test_install_history_no_duplicates
test_install_history_multiple_entries
ASSERT:Multiple entries not recorded expected:<2> but was:<28>
shunit2:ERROR test_install_history_multiple_entries() returned non-zero return code.
perl: warning: Setting locale failed.
perl: warning: Please check that your locale settings:

[Copilot]

@copilot tests are failing

...

Fixed in 85484a6. The test_install_history_multiple_entries test was failing because run cat file | wc -l counted login shell output (MOTD messages from the su - session) along with the actual file content, inflating the line count to 28. Switched to run grep -cF '' file which counts lines within the su - context and only reports the file's line count.

[crramirez]

@copilot shunit2:ERROR test_install_history_multiple_entries() returned non-zero return code.
perl: warning: Setting locale failed.

[Copilot]

@copilot shunit2:ERROR test_install_history_multiple_entries() returned non-zero return code.
perl: warning: Setting locale failed.

Fixed in fc4b0c1. The root cause was that the run function uses su - which starts a login shell, and after MOTD settings are changed, login shell output contaminates stdout. The previous grep -cF '' approach still counted all output lines (including MOTD noise).

Rewrote the tests to use run cat file | grep -cxF 'exact entry' — piping run output through grep with a specific pattern outside the su - context. This follows the same pattern used throughout the test suite (e.g., $(run npm --version | grep -c '11') in nodejs tests, $(run cat ~/.bashrc | grep -c 'PYENV_ROOT') in pythonpi tests). The MOTD output won't match the specific history entries like SETTINGS MOTD MOTD_NEVER, so only actual file content is counted.