tests/run-make: failing tests get incorrectly ignored on rerun

[jieyouxu]

First run (fails on native windows because obviously make not found):

running 362 tests
iii
[run-make] tests\run-make\cat-and-grep-sanity-check ... F
.iii.......ii..iii...........i...ii....i
[run-make] tests\run-make\emit-to-stdout ... F
....i........i..........ii...ii.ii.i.......  88/362
i.......ii..i...i....ii...i.....i..i....ii......ii.......i.i......i......i.....i......i. 176/362
..i....iii.iii...........i.i.iii..i....i.iii.i.......i..iiii.....i..................iii. 264/362
......ii.
[run-make] tests\run-make\split-debuginfo ... F
...i........i...........iii.ii.....ii.iiiiiiiiiiii..i..i...................... 352/362
..........

Second run (no changes):

running 362 tests
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii  88/362
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 176/362
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 264/362
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 352/362
iiiiiiiiii

test result: ok. 0 passed; 0 failed; 362 ignored; 0 measured; 0 filtered out; finished in 2.44ms

The same tests do fail with --force-rerun. Something about the ignore logic is incorrect.

We should not be ignoring failing tests.