microMIPS/opcodes: Refer FPRs rather than FCRs with DMTC1

The DMTC1 instruction operates on a floating-point general register as
its second operand, however in the disassembly of the microMIPS encoding
a floating-point control register is shown instead.  This is due to an
incorrect ordering of the two "dmtc1" entries in the opcode table, which
gives precedence to one using the `G' aka coprocessor format over one
using the `S' or floating-point register format.

The coprocessor format, or OP_REG_COPRO, is used so that GAS supports
referring to FPRs by their numbers in assembly, such as $0, $1, etc.
however in the case of CP1/FPU it is also used by the disassembler to
decode those numbers to the names of corresponding control registers.
This in turn causes nonsensical disassembly such as:

	dmtc1	a1,c1_fir

in a reference to $f0.  It has been like this ever since microMIPS ISA
support has been added.

Correct the ordering of the two entries then by swapping them with each
other, making disassembly output consistent with the regular MIPS DMTC1
instruction as well all the remaining CP1 move instructions.  Adjust all
the test cases affected accordingly.

	opcodes/
	* micromips-opc.c (micromips_opcodes): Swap the two "dmtc1"
	entries with each other.

	gas/
	* testsuite/gas/mips/micromips.d: Update disassembly according
	to "dmtc1" entry fix with opcodes.
	* testsuite/gas/mips/micromips-compact.d: Likewise.
	* testsuite/gas/mips/micromips-insn32.d: Likewise.
	* testsuite/gas/mips/micromips-noinsn32.d: Likewise.
	* testsuite/gas/mips/micromips-trap.d: Likewise.
	* testsuite/gas/mips/micromips@isa-override-1.d: Likewise.