Lower tolerance in BAMG to avoid misidentification of overlapping edges

[prj-]

Fix #319

[cmdoug]

Here is a test case. See the chevalier_etal_2024 example.

[prj-]

Here is a test case. See the chevalier_etal_2024 example.

On my end, this test case is failing with:

$ FreeFem++-mpi -v 0 importgmsh.md -gmshdir examples/chevalier_etal_2024 -dir $workdir -mi nozzle_lg.geo
  Loading 'nozzle_lg.geo' in 'examples/chevalier_etal_2024/'.
  Calling GMSH to build .msh file from .geo file.
  current line = 29 mpirank 0 / 1
Assertion fail : (area > 0)
	line :303, in file ./include/fem.hpp
Assertion fail : (area > 0)
	line :303, in file ./include/fem.hpp
 err code 6 ,  mpirank 0

But that's unrelated to BAMG (and the present PR), isn't it?

[cmdoug]

Sorry about that, tried to rush something to you. I am heading out on vacation for the next week, but will follow up later this month when I have time to look more closely.