Addendum to #614 (fix conditionally allocated array issue)

[grantfirl]

• change 'htrlwu' variable to assumed-shape since it is conditionally allocated to avoid seg faults with GNU/release in SCM

Fixes #651

[climbfuji]

It seems that PR #611 did not capture all arrays with explicit dimensions, please change all explicit dimensions in that file (and any others you come across) to assumed sizes, not just the one array.

[grantfirl]

It seems that PR #611 did not capture all arrays with explicit dimensions, please change all explicit dimensions in that file (and any others you come across) to assumed sizes, not just the one array.

@climbfuji Done. I didn't do a systematic check to see if other files were missing the assumed-shape changes, since that was beyond the scope of this PR (to fix seg faults with the GNU compiler in SCM).

[climbfuji]

Thanks, @grantfirl, for making the additional changes - looks good. Will approve when the regression tests have been done with the UFS.

[grantfirl]

OK, I hope the UFS tests pass. It compiled/ran OK in the SCM at least.

[climbfuji]

I know why the parser didn't catch it, it first looks for invalid type definitions in each file (they are not invalid, just something isn't as capgen wants it), and if this test fails it doesn't parse the dimensions. There might be a few RRTMGP files that need checking, as well as GFS_debug.F90. I'll look at those.