Add unit test for StraightSkeleton robustness

[strk]

A fix for robustness issue reported in #89 was fixed with pull request #92 but no test came with the fix due to lack of WKB input (#94). This ticket is a reminder for adding the test when possible.

[danielcu888]

@strk so this is the ticket where I meant to place that comment...can we accomplish the level of precision testing you require for the relevant module with WKT inputs in fractional form?

[strk]

@danielcu888 I can't tell, possibly, you can try reverting the fix and triggering the bug with fractional WKT input.

[danielcu888]

Sure, I'll try that @strk.

@lbartoletti can I please be assigned to this whilst I am investigating what @strk has proposed?

[danielcu888]

Thanks @lbartoletti this is now in progress.

[danielcu888]

@strk I used the HEXWKB input of the simplified example that you posted in https://github.com/Oslandia/SFCGAL/issues/89 to generate high precision WKT:

POLYGON((1687891.0397187702 5401915.747313647,1684030.8100260005 5402682.520688001,1684000.000000001 5402672.313998163,1681298.834868019 5401963.620584738,1681281.1342950005 5403807.881624001,1683915.838219001 5402676.741864,1684000.000000001 5402711.41802501,1684020.6181620006 5402719.9130770005,1687892.1109056068 5404508.535615864,1687891.0397187702 5401915.747313647))

using the online converter: https://rodic.fr/blog/online-conversion-between-geometric-formats/

and generated the following results:

which looks correct to me. Does this look correct to you? If so, then there is not much more that can be done since I can't seem to reproduce the issue (using high precision WKT tests).