power: qpnp-fg-gen4: Fix the condition for capacity-learning-update

Weighted capacity learning relies on input status to complete the
capacity learning event. The existing logic only relies on
charging_status change to send a cap-learning update, which in a
corner condition may not send an update when input changes status.
Fix this by always sending a capacity learning update irrespective
of the change in input/charge status.

Change-Id: Idfed395e1acad2b3b5c7b19ef137063d186af40f
Signed-off-by: Kiran Gunda <kgunda@codeaurora.org>