Move six dead code stubs to cap/disabled (#101)

ACESII, GAMESS(US), Gaussian-92, Gaussian-94, HONDO and MOLPRO.

Checked rather than assumed, per code: no <InputGenerator>, no
reference from either ResourceDescriptor file, no dialog. So none could
ever set up or submit a calculation, and none was reachable from any
"New ... Calculation" menu. Gaussian-94 carries a <Template>, but with
no generator to fill it that changes nothing.

NWChemMD looked like a seventh candidate on the same "no generator"
test and is deliberately KEPT: it is referenced by BOTH descriptor
files (the nwchem_md_study path the MD tools use), so moving it would
have broken a live feature. Noted in the disabled/README so the next
person does not repeat the inference.

tests/dialogs goes from 21 declared codes to 15 and still passes -- its
coverage is derived from the .edml files, so this is reflected
automatically with nothing to update.

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

Author: andy

data/client/cap/ACESII.edml data/client/cap/disabled/ACESII.edmldata/client/cap/ACESII.edml renamed to data/client/cap/disabled/ACESII.edml

@@ -25,3 +25,26 @@ See data/client/config/disabled-codes-archive.txt for the separate,
 larger removal of AMICA/Gaussian-03/Gaussian-98/MetaDyn's *menu*
 registrations (those codes were fully wired and had real menu entries,
 unlike these four -- a different, easier-to-reverse kind of removal).
+
+
+Second batch, moved 2026-09-22
+------------------------------
+
+ACESII, GAMESS(US), Gaussian-92, Gaussian-94, HONDO and MOLPRO.
+
+Same situation as the original four, and checked the same way before
+moving: none has an <InputGenerator>, none is referenced by
+ResourceDescriptor.xml or ResourceDescriptorRxn.xml, and none had a
+dialog -- so none of them could ever set up or submit a calculation, and
+none was reachable from any "New ... Calculation" menu. Gaussian-94 does
+carry a <Template>, but with no generator to fill it that changes
+nothing.
+
+NWChemMD was considered and deliberately KEPT in data/client/cap: unlike
+these, it IS referenced by both ResourceDescriptor files (the
+nwchem_md_study path used by the MD tools), so moving it would break a
+live feature. Worth re-checking that before moving anything else out of
+here on the assumption that "no generator" means "unused".
+
+Related: #101 collects the remaining retirement candidates, and #56/#57
+(ACESII, HONDO) and #63 (HONDOPLUS) cover specific codes here.