Fix seis/utilmeca

[Chuan1937]

Fixes a numerical instability in meca_axe2dc when P.dip ≈ T.dip.

This issue caused the im flag (fault type) to flip unpredictably between 1 and -1, which in turn shifted the rake by 180 degrees. This resulted in swapped P/T quadrants for psmeca -Sy.

This patch introduces an SEIS_EPSILON tolerance. When |P.dip - T.dip| is within this tolerance, a deterministic im value (-1) is assigned, eliminating the instability.

Issues: #8009

[joa-quim]

Looks good and rely on your testings to considered this as a good fix.

[seisman]

Changes here look reasonable but break existing tests:

	1197 - test/seis/full_moment_tensor.sh (Failed)
	1198 - test/seis/meca_-Smdz.sh (Failed)
	1208 - test/seis/seis_07.sh (Failed)

We need to revert the changes and find the cause.

[Chuan1937]

Changes here look reasonable but break existing tests:

	1197 - test/seis/full_moment_tensor.sh (Failed)
	1198 - test/seis/meca_-Smdz.sh (Failed)
	1208 - test/seis/seis_07.sh (Failed)

We need to revert the changes and find the cause.

I'm trying