rx: Treat scalars larger than 8 bytes as aggregates in rx_push_dummy_call.

rx: Treat scalars larger than 8 bytes as aggregates in rx_push_dummy_call.

This patch fixes the following failures (which are also GDB internal errors)
for the -m64bit-doubles multilib:

FAIL: gdb.base/callfuncs.exp: p t_double_complex_values(dc1, dc2) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: p t_double_complex_values(dc3, dc4) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: p t_double_complex_many_args(dc1, dc2, dc3, dc4, dc1, dc2, dc3, dc4, dc1, dc2, dc3, dc4, dc1, dc2, dc3, dc4) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: p t_double_complex_many_args(dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: p t_long_double_complex_values(ldc1, ldc2) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: p t_long_double_complex_values(ldc3, ldc4) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: p t_long_double_complex_many_args(ldc1, ldc2, ldc3, ldc4, ldc1, ldc2, ldc3, ldc4, ldc1, ldc2, ldc3, ldc4, ldc1, ldc2, ldc3, ldc4) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: p t_long_double_complex_many_args(ldc1, ldc1, ldc1, ldc1, ldc1, ldc1, ldc1,ldc1, ldc1, ldc1, ldc1, ldc1, ldc1, ldc1, ldc1, ldc1) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: noproto: p t_double_complex_values(dc1, dc2) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: noproto: p t_double_complex_values(dc3, dc4) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: noproto: p t_double_complex_many_args(dc1, dc2, dc3, dc4, dc1, dc2, dc3, dc4, dc1, dc2, dc3, dc4, dc1, dc2, dc3, dc4) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: noproto: p t_double_complex_many_args(dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1, dc1) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: noproto: p t_long_double_complex_values(ldc1, ldc2) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: noproto: p t_long_double_complex_values(ldc3, ldc4) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: noproto: p t_long_double_complex_many_args(ldc1, ldc2, ldc3, ldc4, ldc1, ldc2, ldc3, ldc4, ldc1, ldc2, ldc3, ldc4, ldc1, ldc2, ldc3, ldc4) (GDB internal error)
FAIL: gdb.base/callfuncs.exp: noproto: p t_long_double_complex_many_args(ldc1, ldc1, ldc1, ldc1, ldc1, ldc1, ldc1,ldc1, ldc1, ldc1, ldc1, ldc1, ldc1, ldc1, ldc1, ldc1) (GDB internal error)
FAIL: gdb.base/varargs.exp: print find_max_double_real(4, dc1, dc2, dc3, dc4) (GDB internal error)
FAIL: gdb.base/varargs.exp: print find_max_long_double_real(4, ldc1, ldc2, ldc3, ldc4) (GDB internal error)

The assertion failure which is tripped is:

		  gdb_assert (arg_size <= 4);

While it may seem that the patch ought to disallow scalars larger than
4, scalars of size 8 are explicitly handled by the code elsewhere.

This came up because gcc has a complex type that is 16 bytes in length
when 64-bit doubles are used.

gdb/ChangeLog:

	* rx-tdep.c (rx_push_dummy_call): Treat scalars larger than 8
	bytes as aggregates.