Beam beam special case to be handled

[giadarol]

Found in pysixtrack tests that beam beam has a division by zero if:

• beam is round
• particle is exactly in the origin

Fixed in pysixtrack, implement also in Sixtracklib

[martinschwinzerl]

@giadarol : would it be acceptable to use

(x*x + y*y) < ( min_sigma_diff * min_sigma_diff)

instead of

(fabs(x) + fabs(y)) < min_sigma_diff

This would be easier to vectorize and we can reuse the x*x + y*y term multiple times in the same function. Plus, if I understand correctly, this is condition gauges the "real" (euclidian) distance from the reference particle rather than the maximum norm

[giadarol]

That’s perfectly fine