sh: shared register saving code for sh3/sh4/sh4a

sh: shared register saving code for sh3/sh4/sh4a

This patch reworks the sh3/sh4/sh4a register saving code in
the following ways:
 - break out prepare_stack_save_dsp() from handle_exception()
 - break out save_regs() from handle_exception()
 - the register saving order is unchanged
 - align new functions to fit in cache lines
 - separate exception code from interrupt code
 - keep main code flow in a single cache line per exception vector
 - use bsr/rts for regular functions (save pr first)
 - keep data in one shared cache line (exception_data)
 - document the functions
 - tie in the hp6xx code

Signed-off-by: Magnus Damm <damm@igel.co.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>