cmd/internal/gc: avoid turning 'x = f()' into 'tmp = f(); x = tmp' for simple x

cmd/internal/gc: avoid turning 'x = f()' into 'tmp = f(); x = tmp' for simple x

This slows down more things than I expected, but it also speeds things up,
and it reduces stack frame sizes and the load on the optimizer, so it's still
likely a net win.

name                                    old mean                new mean        delta
BenchmarkBinaryTree17              13.2s × (0.98,1.03)     13.2s × (0.98,1.02)  ~ (p=0.795)
BenchmarkFannkuch11                4.41s × (1.00,1.00)     4.45s × (0.99,1.01)  +0.88% (p=0.000)
BenchmarkFmtFprintfEmpty          86.4ns × (0.99,1.01)    90.1ns × (0.95,1.05)  +4.31% (p=0.000)
BenchmarkFmtFprintfString          318ns × (0.96,1.07)     337ns × (0.98,1.03)  +6.05% (p=0.000)
BenchmarkFmtFprintfInt             332ns × (0.97,1.04)     320ns × (0.97,1.02)  -3.42% (p=0.000)
BenchmarkFmtFprintfIntInt          562ns × (0.96,1.04)     574ns × (0.96,1.06)  +2.00% (p=0.013)
BenchmarkFmtFprintfPrefixedInt     442ns × (0.96,1.06)     450ns × (0.97,1.05)  +1.73% (p=0.039)
BenchmarkFmtFprintfFloat           640ns × (0.99,1.02)     659ns × (0.99,1.03)  +3.01% (p=0.000)
BenchmarkFmtManyArgs              2.19µs × (0.97,1.06)    2.21µs × (0.98,1.02)  ~ (p=0.104)
BenchmarkGobDecode                20.0ms × (0.98,1.03)    19.7ms × (0.97,1.04)  -1.35% (p=0.035)
BenchmarkGobEncode                17.8ms × (0.96,1.04)    18.0ms × (0.96,1.06)  ~ (p=0.131)
BenchmarkGzip                      653ms × (0.99,1.02)     652ms × (0.99,1.01)  ~ (p=0.572)
BenchmarkGunzip                    143ms × (0.99,1.02)     142ms × (1.00,1.01)  -0.52% (p=0.005)
BenchmarkHTTPClientServer          110µs × (0.98,1.03)     108µs × (0.99,1.02)  -1.90% (p=0.000)
BenchmarkJSONEncode               40.0ms × (0.98,1.05)    41.5ms × (0.97,1.06)  +3.89% (p=0.000)
BenchmarkJSONDecode                118ms × (0.99,1.01)     118ms × (0.98,1.01)  +0.69% (p=0.010)
BenchmarkMandelbrot200            6.03ms × (1.00,1.01)    6.03ms × (1.00,1.01)  ~ (p=0.924)
BenchmarkGoParse                  8.43ms × (0.92,1.11)    8.56ms × (0.93,1.05)  ~ (p=0.242)
BenchmarkRegexpMatchEasy0_32       180ns × (0.91,1.07)     163ns × (1.00,1.00)  -9.33% (p=0.000)
BenchmarkRegexpMatchEasy0_1K       550ns × (0.98,1.02)     558ns × (0.99,1.01)  +1.44% (p=0.000)
BenchmarkRegexpMatchEasy1_32       152ns × (0.94,1.05)     139ns × (0.98,1.02)  -8.51% (p=0.000)
BenchmarkRegexpMatchEasy1_1K       909ns × (0.98,1.06)     868ns × (0.99,1.02)  -4.52% (p=0.000)
BenchmarkRegexpMatchMedium_32      262ns × (0.97,1.03)     253ns × (0.99,1.02)  -3.31% (p=0.000)
BenchmarkRegexpMatchMedium_1K     73.8µs × (0.98,1.04)    72.7µs × (1.00,1.01)  -1.61% (p=0.001)
BenchmarkRegexpMatchHard_32       3.87µs × (0.99,1.02)    3.87µs × (1.00,1.01)  ~ (p=0.791)
BenchmarkRegexpMatchHard_1K        118µs × (0.98,1.04)     117µs × (0.99,1.02)  ~ (p=0.110)
BenchmarkRevcomp                   1.00s × (0.94,1.10)     0.99s × (0.94,1.09)  ~ (p=0.433)
BenchmarkTemplate                  140ms × (0.97,1.04)     140ms × (0.99,1.01)  ~ (p=0.303)
BenchmarkTimeParse                 622ns × (0.99,1.02)     625ns × (0.99,1.01)  +0.51% (p=0.001)
BenchmarkTimeFormat                731ns × (0.98,1.04)     719ns × (0.99,1.01)  -1.66% (p=0.000)

Change-Id: Ibc3edb59a178adafda50156f46a341f69a17d83f
Reviewed-on: https://go-review.googlesource.com/9721
Reviewed-by: David Chase <drchase@google.com>