BUG: inconsistent meaning of skip_covered between coverage report and coverage html

[neutrinoceros]

Describe the bug

The "total" line in coverage html's output isn't consistent with that from coverage report when [report] skip_covered=true, despite the parameter being explicitly used by both commands.
Basically, the inline report includes 100% covered files in the total metric, while the html report doesn't, effectively giving a different metric.

To Reproduce
I'll link a github repo with a full minimal reproducer once this issue is published and has an associated number.

update: here it is https://github.com/neutrinoceros/reprod_coveragepy_1784

Expected behavior
I don't think any of these metrics are incorrect, and in fact they're both useful, but the inconsistency is confusing, so I'd suggest including both metrics in all reports when [report] skip_covered=true, possibly only when they are different (which should be anytime some but not all files are at 100% coverage).

Additional context
Add any other context about the problem here.