docs: correct function and statistic names the modules do not export

[denys-i-didww]

Summary

#4241 corrected five parameter names the modules do not export, and its Scope section
said the same class in the Exported Functions, MI Commands and Statistics sections was
left for a separate change. This is that change. Three names in three modules; the code is
right in each and the page is wrong.

Module	Section	README documents	Code exports
janus	Exported Functions	janus_send_requeest()	janus_send_request — janus_mod.c:67
event_sqs	Exported Functions	sqs_publish_message()	sqs_send_message — event_sqs.c:65
registrar	Exported Statistics	defaults_expires	default_expire — reg_mod.c:233

A script calling a function the module does not export does not load, and an MI query for
a statistic that does not exist returns Statistics Not Found.

Details

janus is a plain misspelling — requeest — and only in the heading, README.md:163.
Two lines below, the example's title and the example itself already read
janus_send_request(), so the page contradicts itself and the heading is the odd one out.

event_sqs documents the name of the C function rather than the name of the script
command. event_sqs.c:65 reads

{"sqs_send_message", (cmd_function)sqs_publish_message, {

so sqs_publish_message is the static implementation (:48, :425) and
sqs_send_message is what a script may call. The README uses the implementation's name in
all four places — the prose at :77, the heading at :112, the example title at :130
and the example line at :134 — and nothing on the page corrects the reader.

registrar documents the statistic as defaults_expires, while stat_export_t
declares default_expire. The documented name is one letter away from the parameter
default_expires, which does exist (reg_mod.c:203), and the description under the
heading — "The value of default_expires parameter" — correctly describes that parameter;
the statistic does mirror it (reg_mod.c:405). Only the heading changes. The module's
other four documented statistics match stat_export_t exactly and are left alone.

Reproduced

opensips 4.1.0-dev, Debian 12, built from source.

The function name fails when the script is loaded, not at modparam parsing — a different
failure point from #4241, which is part of why this is a separate PR:

$ opensips -C -f j.cfg            # janus_send_requeest("a","b"); as the page reads today
CRITICAL:core:yyerror: parse error in j.cfg:9:33-34: unknown command <janus_send_requeest>, missing loadmodule?

$ opensips -C -f j.cfg            # janus_send_request("a","b"); as this PR documents
NOTICE:core:main: config file ok, exiting...

The statistic fails at the MI query, a third failure point. registrar running with
usrloc, signaling, tm and mi_http:

$ ... "method":"get_statistics","params":{"statistics":["defaults_expires"]}
{"jsonrpc":"2.0","error":{"code":404,"message":"Statistics Not Found"},"id":1}

$ ... "method":"get_statistics","params":{"statistics":["default_expire"]}
{"jsonrpc":"2.0","result":{"registrar:default_expire":3600},"id":1}

Listing every statistic on that instance returns registrar:default_expire and no
defaults_expires.

event_sqs is the one row not run. It links against libaws-cpp-sdk-sqs, which Debian
does not package, so the module cannot be built without building the AWS SDK first.
Compiling event_sqs.c on its own does work, and nm on the object shows
sqs_publish_message as a local text symbol — the static function — while
sqs_send_message is the name in cmds[]. That is source-level evidence, not a run, and
I would rather say so than imply otherwise.

Scope

Three renames, nothing else. The same sweep, re-run against 08e7f1c over every
module's Exported Functions, Asynchronous Functions, MI Functions, Statistics and
Pseudo-Variables sections, turns up two more documented names — presence_callinfo's
sca_set_calling_line() and ratelimit's rl_bin_status — that are not renames at all:
neither has an implementation anywhere in the tree, so there is nothing to point the
documentation at and the fix is yours to choose. They need an issue, not a patch, and are
not in this PR.

Everything else that looked like a mismatch is not one, and is recorded here so it is not
re-reported: MI command names come from #defines (usrloc, rtpengine, pike and
others), callops's E_CALL_TRANSFER and E_CALL_HOLD are built by token pasting from
"E_CALL_" #_name, registrar and mid_registrar's pn_process_purr() comes from a
macro in lib/reg, and call_center, janus and rr have #### sub-headings where
their siblings use ###, which a heading scan reads as names.

Compatibility

Documentation only, no behaviour change. Nothing that works today stops working: none of
the three documented names could ever have been used.

---

AI assistance disclosure. The defect list and the wording of this change were produced
with AI assistance. Every item was checked against the source before submission:

• each exported name is the literal in the module's own export array — janus_mod.c:67
  (cmd_export_t), event_sqs.c:65 (cmd_export_t), reg_mod.c:233 (stat_export_t);
• each documented name appears nowhere in the tree as a string literal — repository-wide
  grep over *.c, *.h and *.cpp, which is what distinguishes these three from the
  macro-defined names listed under Scope;
• every occurrence in each README was changed, so no page is left carrying both spellings
  — one in janus, four in event_sqs, one in registrar;
• registrar's other four statistics were checked against stat_export_t and left alone,
  and the parameter default_expires it is confused with is real — reg_mod.c:203;
• the two failures and the two successes were run, not inferred — the output is quoted
  above, on a build of this branch;
• event_sqs was not run, and the reason and the weaker evidence are stated in place;
• the two names with no implementation were confirmed absent by the same repository-wide
  grep before being excluded from this PR.