fix(processlist): index io_counters instead of testing membership in it - #3675

The defect

_sum_stats(key, sub_key) (glances/plugins/processlist/__init__.py:1030) guards the accumulation with a mapping test:

elif sub_key in p[key]:
    ret += p[key][sub_key]

sub_key is only a mapping key for memory_info. For io_counters it is an index — the value is the list [read_bytes, write_bytes, read_bytes_old, write_bytes_old, io_tag] (glances/processes.py:553) — so 0 in p['io_counters'] asks whether the counters contain the number zero.

Measured on develop:

stats = [[5000, 700, 2000, 300, 1],
         [3000, 400, 1000, 200, 1]]

_sum_stats('io_counters', 0)          ->    0     expected 8000
_sum_stats('io_counters', sub_key=2)  ->    0     expected 3000
_sum_stats('memory_info', 'rss')      ->   40     correct — it is a mapping

stats = [[5000, 700, 0, 300, 1]]
_sum_stats('io_counters', 0)          -> 5000     summed only because 0 is in the list

The symptom

The R/s and W/s figures on the process-list summary row — _msg_curse_sum, rendered whenever a process filter is active, for current/min/max — are computed as _sum_stats(…, 0) - _sum_stats(…, sub_key=2) at :945 and :955. Both terms come back 0 for every process actually doing IO, so the row reads zero; and when some idle process happens to carry the literal index in its counters, it reads a number that belongs to nobody. The VIRT/RES totals on the same row are right, which is what makes the zero look plausible.

The fix

Branch on the container type, and bounds-check the index so a short list is skipped rather than raising IndexError. The mapping path is untouched.

Verification

Windows 11, Python 3.14.7.

tests/test_core.py    47 passed, 6 skipped     (46 before — the case added)
ruff check            All checks passed

Mutation-checked, after confirming the edit applied: disabling the new branch fails the new case.

Tests

tests/test_core.py::test_010c_processlist_sum_stats_indexes_lists — all four io_counters indexes against known values, the mapping path unchanged (including an absent key), the "list contains the literal index" case that must no longer decide anything, and an out-of-range index that must be skipped rather than raise.

Note

The same shape exists at glances/folder_list.py:122,135 (i in self.timer_folders, an int against a list of Timers, so the per-folder folder_N_refresh option never takes effect). I left it out of this PR since it is a different plugin with a different symptom — happy to send it separately if useful.