fix: replace fragile floating-point equality comparisons

- m_ibm.fpp: Re(1) /= 0._wp -> Re(1) > 0._wp (matches convention in
  m_global_parameters.fpp where Re is checked with > 0)
- m_ibm.fpp: sqrt(sum(axis**2)) == 0 -> < sgm_eps (floating-point
  magnitude should not be compared with exact equality)
- ExtrusionHardcodedIC.fpp: dummy_x == x0 -> f_approx_equal (coordinates
  read from files should not be compared with exact equality)

Author: sbryngelson

src/common/include/ExtrusionHardcodedIC.fpp

@@ -109,7 +109,7 @@
             do
                 read (unit2, *, iostat=ios2) dummy_x, dummy_y, dummy_z
                 if (ios2 /= 0) exit
-                if (dummy_x == x0 .and. dummy_y /= y0) then
+                if (f_approx_equal(dummy_x, x0) .and. (.not. f_approx_equal(dummy_y, y0))) then
                     yRows = yRows + 1
                 else
                     exit

src/simulation/m_ibm.fpp

@@ -904,7 +904,7 @@ contains
 
         if (viscous) then
             do fluid_idx = 1, num_fluids
-                if (fluid_pp(fluid_idx)%Re(1) /= 0._wp) then
+                if (fluid_pp(fluid_idx)%Re(1) > 0._wp) then
                     dynamic_viscosities(fluid_idx) = 1._wp/fluid_pp(fluid_idx)%Re(1)
                 else
                     dynamic_viscosities(fluid_idx) = 0._wp
@@ -1110,7 +1110,7 @@ contains
 
         if (p == 0) then
             normal_axis = [0, 0, 1]
-        else if (sqrt(sum(axis**2)) == 0) then
+        else if (sqrt(sum(axis**2)) < sgm_eps) then
             ! if the object is not actually rotating at this time, return a dummy value and exit
             patch_ib(ib_marker)%moment = 1._wp
             return