Optimize BE load and store of uint256

[chfast]

This is critical for EVM MSTORE and MLOAD. GCC was not properly optimizing these because of the https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107837 bug.

load_store_be<uint128>_mean                  +0.0115         +0.0114             2             2             2             2
load_store_be<uint256>_mean                  -0.8506         -0.8506            23             3            23             3
load_store_be<uint512>_mean                  -0.4334         -0.4335            40            23            40            23

[codecov-commenter]

Codecov Report

Merging #281 (857bfe6) into master (15e3a02) will not change coverage.
The diff coverage is 100.00%.

❗ Current head 857bfe6 differs from pull request most recent head b590752. Consider uploading reports for the commit b590752 to get more accurate results

@@            Coverage Diff            @@
##            master      #281   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           10        10           
  Lines         1941      1951   +10     
=========================================
+ Hits          1941      1951   +10     

Flag	Coverage Δ
32bit	100.00% <100.00%> (ø)
gcc	99.48% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
include/intx/intx.hpp	100.00% <100.00%> (ø)
test/unittests/test_intx.cpp	100.00% <100.00%> (ø)