[trivial fix] Fix build when ccd_real_t == float

[v4hn]

libccd supports single or double precision builds.
It seems like the developers build it with double precision.
But if you actually require double precision you should check for it in the configuration.

Without the patch the build fails because stl does not support std::max(double, float) (problems with the initializer_list overload being preferred).
A fact that by itself is rather frustrating as well...

---

This change is 

[sherm1]

Thanks @v4hn! pending minor cosmetic fix (line length)

Reviewed 1 of 1 files at r1.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @v4hn)

---

include/fcl/narrowphase/detail/convexity_based_algorithm/gjk_libccd-inl.h, line 1696 at r1 (raw file):

  const ccd_real_t v0_dist =
      std::sqrt(ccdVec3Len2(&nearest_edge->vertex[0]->v.v));
  const ccd_real_t plane_threshold = kEps * std::max(static_cast<ccd_real_t>(1.0), v0_dist);

nit: please keep line length to <= 80

[v4hn]

I just pushed an additional line break.

Sorry, I'm not sure whether I'm expected to do anything at reviewable because the page keeps loading forever saying "Resuming session". ❓

[sherm1]

In Reviewable if the reviewer's comment begins with "nit" or "minor" you can resolve it yourself by answering with a comment beginning with "done" (meaning: you fixed it) or "ok" (meaning: I thought about it but am going to leave it this way for [reason]).

But it's all good now -- thanks!

Reviewed 1 of 1 files at r2.
Reviewable status: complete! all files reviewed, all discussions resolved