ai.nwchem: remove a die() that broke input generation for ordinary jobs

Reported from a live session: NWChem calculations could not be saved --
"Input files could not be generated -- input parsing command
/opt/ecce/scripts/parsers/ai.nwchem ... failed". Gaussian was fine.

sub CCThresh died when CC.ConvergenceEnergy.Value was zero or absent.
nwch.tpl carries ##CCThresh## whatever the theory is, so any job with
no coupled-cluster settings reached it -- including a plain SCF energy,
which is about the most common NWChem job there is.

The die also contradicted the three lines directly beneath it, which
already handle an absent threshold by emitting nothing.

It did no visible harm while the template engine swallowed die()s: the
tag's line silently vanished and generation carried on with a quietly
incomplete deck. a6d0512 (#92) made those errors real, which turned a
silent wrong answer into a hard failure. That is the fix working as
intended -- CLAUDE.md predicted it would "surface errors that were
previously invisible" -- but this particular error was never worth
raising.

Confirmed NOT a regression from this session's changes: the version of
ai.nwchem from before any of them fails identically (exit 255, same
message), so the die is what is wrong. Water/STO-3G/RHF now generates a
valid deck and exits 0.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EFMUNbw2UpRPmgsqviAowb

Author: andy

scripts/parsers/ai.nwchem

@@ -953,8 +953,20 @@ sub CCThresh {
   local($thresh,$result);
 
   $thresh = $AbiDict{"CC.ConvergenceEnergy.Value"};
-  die "ai.nwchem:  AbiDict{\"CC.ConvergenceEnergy.Value\"} is zero\n"
-      if ($thresh == 0.0);
+  # There used to be a die() here when the threshold was zero or absent.
+  # It contradicted the three lines below, which already handle that case
+  # by emitting nothing -- and nwch.tpl carries ##CCThresh## whatever the
+  # theory is, so ANY job without coupled-cluster settings (an ordinary
+  # SCF energy, say) reached it.
+  #
+  # It did no visible harm while the template engine swallowed die()s
+  # (#92): the tag's line silently vanished and generation continued.
+  # Since a6d0512 made those errors real, this aborted NWChem input
+  # generation outright -- "Input files could not be generated ...
+  # ai.nwchem ... failed" -- for the most common kind of NWChem job there
+  # is. Reported from a live session 2026-09-22; confirmed identical
+  # before and after that day's changes, so it is the die that is wrong,
+  # not anything recent.
   $result = "";
 
   if ($thresh) {