COPY from previous stage always misses

[ashlynrlee]

Bug Copy instructions from previous stages are not cached correctly.

Steps to reproduce:

$pwd
 ~/charliecloud/examples
$ ch-image pull ubuntu:20.04
$ ch-image build --force -t nvidia -f Dockerfile.nvidia .
$ ch-image build --force -t nvidia -f Dockerfile.nvidia .

Output:

39. RUN.F make
copying image ...
 43* FROM ubuntu:20.04
 45. COPY  --from=0['/usr/local/cuda-11.2/samples/0_Simple/matrixMulCUBLAS'] -> '/'
copying image ...
will use --force: debderiv: Debian 9+, Ubuntu 14+, or other derivative
 51. COPY  --from=0['/usr/local/cuda-11.2/targets/x86_64-linux/lib/libcublas.so.11.4.1.1043'] -> '/usr/local/lib'
 52. COPY  --from=0['/usr/local/cuda-11.2/targets/x86_64-linux/lib/libcublasLt.so.11.4.1.1043'] -> '/usr/local/lib'
 53. RUN.F ldconfig
 warning: --force specified, but nothing to do
grown in 13 instructions: nvidia

These should have been cache hits instead full Copy instructions.
Notice 43* where the asterisk indicates a cache hit and 45 where no asterisk shows the first cache miss.