FDS Source: Issue #12019. Fix bug related to double precision reals

[mcgratta]

No description provided.

[mcgratta]

This PR does the following:

1. Change the name ALMOST_ONE to ONE_M_EPS
2. Add ONE_P_EPS=1+EPSILON
3. Fix bug in read.f90 related to finding bounds for spatial integration.

There are a few places in the code where we want a number to be a little bigger or smaller to avoid computer round-off. Previously, we would typically write Y - TWO_EPSILON_EB to be a number just less than Y if it were 12.00000000 and we wanted to round down to 11.999999. Like when converting XB on the DEVC line to find the proper cells over which to integrate a spatial stat. Anyway, EPSILON(X) is the smallest number for which 1-EPSILON(X)<1 . But if we add or subtract one or two epsilons from a larger number, the result may not be larger or smaller than the number. Instead, it is better to multiply our number by 1-EPS or 1+EPS to ensure that the result will be significant.

Note, this does not affect the vast majority of uses for TWO_EPSILON_EB where we just want to know if a number is close to zero.