relax unit test thresholds to fix 32bit issues

[jlblancoc]

This will solve part of the pending issues in #1605 :

(cc @dkogan )

• i386 fails one of the first set of tests: it looks like it just barely misses some thresholds, so it's probably not a failure. But since it failed something in the first set, we don't know if any of the second set of tests would have failed

[varunagrawal]

Would it be wise to set the thresholds based on preprocessor flags indicating the architecture type?

If we have the flag GTSAM_ARCH_TYPE, we could check at compile time and set this accordingly. Just a suggestion.

[ProfFan]

@jlblancoc Before merging could you add a constexpr if check on the current architecture?

[jlblancoc]

Take a look at the patch now.

I checked that i686-linux-gnu-g++ defines both preprocessor symbols. i386 is actually obsolete and only i686 will remain for a few years as a main Debian arch, but I added both just in case.

@varunagrawal The idea of a cmake-level architecture variable that maps into a macro seems good (I think that is what your proposal is about?), but if we are using it just here, maybe just the macros are the minimal solution...

[jlblancoc]

One CI task failed, but I think it's not related to this PR's changes... I have re-launched it.

[varunagrawal]

Awesome! Merged