Relying on PM-ops for shutting down PHY clocks was a
bad idea since the users (e.g. PCIe/SATA) might not
have been suspended by then.

The main culprit for not shutting down the clocks was
the stray pm_runtime_get() call in probe.

Fix the whole thing in the right way by getting rid
of that pm_runtime_get() call from probe and
removing all PM-ops. It is the sole responsibility
of the PHY user to properly turn OFF and de-initialize
the PHY as part of its suspend routine.

As PHY core serializes init/exit we don't need
to use a spinlock in this driver. So get rid of it.
Signed-off-by: Roger Quadros <rogerq@ti.com>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>