Make the SolisCloud API call timeout configurable (default unchanged at 10s)

[calindragomir84]

Problem

The timeout for a single SolisCloud API call is hard-coded to 10 seconds, in two places in
soliscloud_api.py (_get_data and the POST helper). When SolisCloud is slow — which it is under
load — calls are cut off before they return, and the integration reports No valid inverters found, login failed / No inverters found. The call is not being refused; it is not returning in time.

Measured today from a machine outside Home Assistant, using plain unsigned POSTs to
/v1/api/userStationList and /v1/api/inverterList (a healthy API rejects these instantly with
403 "Date can not be empty", so the round-trip cost is all server-side):

requests	24
responded	11 — median 9.03 s, max 14.21 s
no response within 15 s	12
502 Bad Gateway	1

Re-scored against the 10 s budget the integration actually allows, only 6 of 24 would have
completed. Five responses arrived between 12.1 s and 14.2 s — the API answered, just too late.

Response-time distribution, seconds: 1.7 · 2.8 · 4.5 · 6.2 · 7.8 · 9.0 · 12.1 · 12.2 · 12.6 · 12.9 · 14.2

Change

Adds a timeout option, exposed in the options flow next to refresh_ok / refresh_nok, and in
the initial setup step.

• const.py — CONF_TIMEOUT, DEFAULT_TIMEOUT = 10
• soliscloud_api.py — SoliscloudConfig takes timeout: int = 10, exposes a timeout property,
  and both async_timeout.timeout(10) call sites become async_timeout.timeout(self.config.timeout)
• __init__.py — reads the option, falls back to DEFAULT_TIMEOUT
• config_flow.py — field in both the options flow and the initial step
• translations/en.json — label

Default is 10, so behaviour is unchanged for anyone who does not set it. The constructor argument
is keyword-with-default, so existing callers keep working.

Testing

Running on Home Assistant 2026.9.1 against a live SolisCloud plant:

• integration loads with no errors; no TypeError/KeyError on any code path exercised
• the new field appears in the options form (integer, default 10) and the saved value persists
• set to 30 and confirmed stored in the config entry

Like the existing refresh_ok / refresh_nok, the value is read when the entry is set up, so a
restart is needed for a change to take effect — consistent with the current
"Some changes require a restart of Home Assistant" note. Happy to add an update listener in a
separate PR if you'd like all three to apply without a restart.

Note

translations/en.json also loses one trailing space on the portal_password line — a side effect of
re-serialising the file, not an intentional change. Say the word and I'll restore it.