Bug#1050236: FTBFS: 30 - testlpe fails

[Sven Mueller <smu@google.com>]

Hi.

Two notes:

1) This doesn't just fail on rscv64 but also on amd64 and i386 with
the same error as in the linked build log from the original report:

[...]
[ RUN      ] LPETest.MeasureSegments_multi_px_1_0_2
add_actions_edit_document: no app!
add_actions_pages: no app!
add_actions_undo: no app!
./testfiles/lpespaths-test.h:96: Failure
The difference between pointc[Geom::X] and pointd[Geom::X] is
1.4044480000000021, which exceeds precission, where
pointc[Geom::X] evaluates to 297.94489599999997,
pointd[Geom::X] evaluates to 299.34934399999997, and
precission evaluates to 1.3999999999999999.
./testfiles/lpespaths-test.h:97: Failure
The difference between pointc[Geom::Y] and pointd[Geom::Y] is
1.5330759999999941, which exceeds precission, where
pointc[Geom::Y] evaluates to 295.84707600000002,
pointd[Geom::Y] evaluates to 297.38015200000001, and
precission evaluates to 1.3999999999999999.
./testfiles/lpespaths-test.h:96: Failure
[...]
[  FAILED  ] LPETest.MeasureSegments_multi_px_1_0_2
[  FAILED  ] LPETest.MeasureSegments_multi_mm_1_0_2

2) If I'm not mistaken, that means that the "forwarded" tag is wrong
here (or at least the linked issue is: This is not segfaulting (like
the "forwarded" bug) but erroring out on imprecise measurements.

If I recall correctly, there is a separate issue on github about this,
somehow related to font rendering(?). However, I can't seem to find it
anymore.

Kind regards,
Sven