Upgrade systeminformation to 5.x, clearing the last advisories

[sachinsachdeva]

systeminformation is the extension's only production dependency and the only one that ships. Every version up to 5.31.6 carries command-injection advisories — eleven of them. Clearing those means 4.x → 5.x, which renames most of the surface this extension reads.

npm audit now reports 0 vulnerabilities, down from 8 (1 critical, 5 high, 1 moderate, 1 low).

Stacks on #6. This branch contains that commit, so merging this alone also lands the dev-dependency patches. Merge #6 first if you'd rather keep them separate.

Renames that would have broken readings, not merely changed them

What	v4	v5	Consequence if missed
CPU frequency	si.cpuCurrentspeed()	si.cpuCurrentSpeed()	Throws. ResMon.update() gathers resources with Promise.all and has no error handling, so the entire status bar freezes
CPU usage	currentload_idle etc.	currentLoadIdle etc.	undefined → usage renders as NaN%
CPU temp	main === -1 when absent	main === null	Check was !== -1, so a Mac with no sensor shows "null C"
Battery	hasbattery, ischarging, …	hasBattery, isCharging, …	Battery section never appears

fsSize() changed meaning, and for the better

use is now used ÷ (used + available) — what df reports — where it was used ÷ size. On APFS those differ sharply, because every volume in a container reports the container's size. v5 also exposes available directly, so free space is no longer computed as size - used:

                         v4 (size - used)   v5 (available)   df
free space on this Mac        37 GB             9.80 GB      9883M
percent used                  16.22%            95.11%       96%

The old figures were wrong, not merely different. All four disk.format modes were checked against df.

Verification

Run on an Apple M4 against system tools:

Reading	Result
CPU usage + user/system split	21.59% (13.94% user, 7.65% system)
Per-core load bars	▄▃▂▂▂▁▂▂▂▂ peak 45% on core 0
CPU frequency	2.40 GHz
CPU temperature	correctly absent (no sensor) rather than null C
Memory / swap	14.16 GB of 16.00 GB (88.48%), swap 6.21 GB of 7.00 GB
Disk, all four formats	match df (4.89% remaining, 9.80 GB free)
Battery	hidden — see caveat below

48 tests pass, npm run lint clean, and the extension was smoke-tested through a mocked vscode module with every section rendering and the host exiting cleanly.

Caveat worth knowing: this machine has no battery, so those code paths could not be exercised at runtime. Instead every property the code reads off battery() was cross-checked against the keys v5 actually returns — all seven present. That is weaker than a live check, and a reviewer on a laptop could confirm it in a moment.

Also corrected: the comment claiming si.cpu() throws on Apple Silicon. It no longer does.

🤖 Generated with Claude Code