[gabrielmancini]

Fixes #2433, following @petermm's review there.

network:sta_rssi/0 sent {self(), Ref, rssi} to network_port and then received anything:

• Other -> {error, Other} took the first message in the caller's mailbox. A gen_server, or any process with other traffic (timers, socket data, '$gen_call's), lost that message and got {error, Message}, while the real reply stayed queued for the next call.
• With no after, the caller waited forever when no reply came, for example when the port went away between whereis/1 and the send.

It now receives only {Ref, {error, _}} and {Ref, {rssi, _}} for its own request, and returns {error, timeout} after 5000 ms. As the review points out, a timeout does not cancel the request, so a late {Ref, Reply} may still reach the caller's mailbox; the edoc says so. The request protocol, the /0 API and the success/error mapping are unchanged, and the drivers are untouched.

Test. tests/libs/eavmlib/test_network.erl registers a fake network_port and checks: network_down with no port; an unrelated message already in the mailbox is left there; a reply with another reference is ignored (and left there); an error reply is passed through; no reply gives {error, timeout}.

• generic_unix, without the fix: test_network fails with {badmatch,{error,unrelated}}; with it, all of test_eavmlib.avm passes.

The timeout case makes test_eavmlib about 5 s longer.

These changes are made under both the "Apache 2.0" and the "GNU Lesser General Public License 2.1 or later" license terms (dual license).

SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later

[gabrielmancini]

Rebased on release-0.7 after #2436 was merged: the conflict was the CHANGELOG only, the code is the approved one.