Automatic merge from submit-queue

Use unsafe operations to get fast memory copy

We can avoid doing allocations by using legal unsafe.Pointer allocations to
assign pointers to equivalent memory locations (for memory identical types).

@wojtek-t

```
benchcmp /tmp/old /tmp/new
benchmark                                      old ns/op     new ns/op     delta
BenchmarkPodConversion-8                       3590          2438          -32.09%
BenchmarkNodeConversion-8                      4774          1773          -62.86%
BenchmarkReplicationControllerConversion-8     3711          2919          -21.34%
BenchmarkEncodeCodecFromInternalProtobuf-8     5692          4816          -15.39%
BenchmarkDecodeCodecToInternalProtobuf-8       7762          6540          -15.74%
BenchmarkEncodeCodecFromInternal-8             32668         31158         -4.62%
BenchmarkDecodeCodec-8                         67260         63348         -5.82%
BenchmarkDecodeIntoInternalCodec-8             62903         64510         +2.55%

benchmark                                      old allocs     new allocs     delta
BenchmarkPodConversion-8                       26             13             -50.00%
BenchmarkNodeConversion-8                      21             9              -57.14%
BenchmarkReplicationControllerConversion-8     23             17             -26.09%
BenchmarkEncodeCodecFromInternalProtobuf-8     16             8              -50.00%
BenchmarkDecodeCodecToInternalProtobuf-8       72             64             -11.11%
BenchmarkEncodeCodecFromInternal-8             68             60             -11.76%
BenchmarkDecodeCodec-8                         268            260            -2.99%
BenchmarkDecodeIntoInternalCodec-8             268            260            -2.99%

benchmark                                      old bytes     new bytes     delta
BenchmarkPodConversion-8                       3324          2377          -28.49%
BenchmarkNodeConversion-8                      4800          1696          -64.67%
BenchmarkReplicationControllerConversion-8     3456          3072          -11.11%
BenchmarkEncodeCodecFromInternalProtobuf-8     3992          3392          -15.03%
BenchmarkDecodeCodecToInternalProtobuf-8       5560          4960          -10.79%
BenchmarkEncodeCodecFromInternal-8             7180          6576          -8.41%
BenchmarkDecodeCodec-8                         13920         13320         -4.31%
BenchmarkDecodeIntoInternalCodec-8             13920         13320         -4.31%
```