Enhancement-653: exit is a second name for quit

Enhancement-653: exit is a second name for quit

Scope: src/frontend/commands.c (an exit entry beside quit in the ngspice and
the nutmeg command tables, both bound to com_quit), src/frontend/com_sweep.c
(sw_analysis_is_destructive: exit in the refused list beside quit),
examples/exitcmd_examples/ (new, 10 checks per solver); handbook
§3.11. ngspice only. Requested by the
user.

Suites: exitcmd_examples 10 of 10 per solver,
both solvers (9 of 10 fail on the E-652 binary); sweepanalysis, carddefault,
probeblock unchanged; full sweep 526 of 526.

What was asked

Whether a .control block may end ngspice with exit. It could not: exit was no
command at all, so

.control
op
exit
print v(out)
.endc

printed exit: no such command available in ngspice and went on with the next line —
in batch mode with everything after it. The command that leaves ngspice is quit. The
user asked for exit as a second name for it.

What changed

• commands.c: an exit entry follows quit in spcp_coms (ngspice) and
  nutcp_coms (nutmeg), with the same handler, argument shape and flags, and the help
  text "Quit ngspice (the same as quit)". Being the same function, exit takes what
  quit takes: an integer is the process's exit status (exit 3), and noask skips
  the "Are you sure you want to quit (yes)?" question that set askquit raises when a
  simulation is still in progress or a plot has not been written. In the shared library
  it returns through controlled_exit(1000 + code) as quit does. Command completion
  and oldhelp exit come from the table entry.
• com_sweep.c: sweep -analysis exit is refused up front, as -analysis quit
  has been since E-341: the analysis a sweep runs per point must
  leave the circuit standing. The script continues after the refusal.
• Nothing else consults the name. The ic.file scan in inpcom.c already ignores a
  control line that begins with a letter it does not examine, and the cp_evloop("quit")
  in main.c is the program's own call. With set unixcom an unknown command is handed
  to the shell; exit is found in the table first, so it ends ngspice rather than a
  child shell.

Verification

check	result
exit in a .control block, batch mode	no "no such command", the lines after it do not run, the done banner, status 0
exit 3 / exit noask	status 3 / status 0
exit 7 inside an if; exit 2 inside a repeat	status 7; one pass, status 2
quit, quit 5	unchanged (0, 5)
pipe mode: exit, exit 4	the session ends before the next line; status 0 / 4
oldhelp exit, oldhelp quit	exit : Quit ngspice (the same as quit).; the quit line unchanged
set askquit, an unsaved plot: exit then yes; exit noask	the question, then the end; no question
sweep … -analysis exit, -analysis quit	both refused ("would destroy the circuit"), the script continues

Full sweep 526 of 526 on both solvers.

What this does not do

• exit is a command, not a shell exit: it ends ngspice (or, in the shared library,
  returns control to the host) exactly as quit does. It is not added to the
  documentation browser's subjects, which quit has none of either.

Write-up: enhancements_doc/Enhancement-653.md