fix(cli): honour --endpoint and AWS_ENDPOINT_URL, and make status work again

[DaviReisVieira]

Fixes #171

Problem

stackport list, describe and status take --endpoint and honour AWS_ENDPOINT_URL, but only on a machine that never opened the web UI. Once ~/.stackport/endpoints.json exists, the saved default wins and both are ignored.

On top of that, stackport status has been failing for everyone: it called _probe_service(svc) without the endpoint_url that helper requires, and exited 2 with "missing 1 required positional argument".

Cause

The commands wrote os.environ["AWS_ENDPOINT_URL"] after backend.config had already read and deleted that variable, so nothing in StackPort read it. get_client() with no endpoint_url resolves through the endpoint store, which loads the saved file. The fresh-machine cases only worked because botocore picks the variable up on its own.

Fix

• One helper resolves the endpoint per command: explicit --endpoint (or AWS_ENDPOINT_URL at startup) wins, otherwise the saved default, the same one the web UI uses
• get_client and _probe_service receive endpoint_url and region explicitly
• The os.environ writes are gone

Verified

Against my own ~/.stackport (saved default local = 4566), with nothing listening on 4599 so the error shows the URL called:

configuration	before	after
AWS_ENDPOINT_URL=...:4599, fresh data dir	4599	4599
AWS_ENDPOINT_URL=...:4599, existing ~/.stackport	4566	4599
--endpoint ...:4599, fresh data dir	4599	4599
--endpoint ...:4599, existing ~/.stackport	4566	4599
no flag, no env, existing ~/.stackport	4566	4566

stackport status --endpoint http://localhost:4566 against MiniStack: 35 services, 35 available.

Four new tests pin the behaviour: flag reaches get_client for list and describe, no flag falls back to the saved default, and status probes every service at the given endpoint. Full backend suite: 544 passed.