Crash test update

[Caolán McNamara]

On Wed, 2020-04-08 at 04:30 +0000, crashtest wrote:
> New crashtest update available at 
> https://dev-builds.libreoffice.org/crashtest/27d04f6dbf38aa28fb7215590d578c4567db57

There's a bunch of the same assert listed there:

vcl/source/outdev/map.cxx:366: long int ImplLogicToPixel(long int, long
int, long int, long int, long int): Assertion `nMapNum == 0 ||
std::abs(n) < std::numeric_limits<long>::max() / nMapNum / nDPI'
failed.

using ./soffice --headless --convert-to pdf on e.g.
the svg attachment 
https://bugzilla.mozilla.org/attachment.cgi?id=9058902 (of mozilla bug
https://bugzilla.mozilla.org/show_bug.cgi?id=1545040) reproduces this

The assert seems to have become noticeable since

commit 62ac8333999c661432adb0a18245a399daa89dcb (HEAD)
Date:   Fri Feb 14 16:47:14 2020 +0100

    tdf#130655 callback interface for 3D and secure dash

But that looks blameless in the sense that it probably just timed out
before that change. The original svg contains massive numbers in the
input source path. Anyone got any thoughts about what we should do with
cases like this ?