clocksource/drivers/tegra: Support per-CPU timers on all Tegra's

Assign TMR1-4 per-CPU core on 32bit Tegra's in a way it is done for
Tegra210. In a result each core can handle its own timer events, less
code is unique to ARM64 and Tegra's clock events driver now has higher
rating on all Tegra's, replacing the ARM's TWD timer which isn't very
accurate due to the clock rate jitter caused by CPU frequency scaling.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Acked-By: Peter De Schrijver <pdeschrijver@nvidia.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>