Supporting consecutive STF_INST_MEM_CONTENT records

[bit-hack]

I have been hitting an issue when trying to encode memory accesses of greater than 8 bytes.

The specification says of STF_INST_MEM_CONTENT records:

For memory accesses with data size greater than 8 bytes, use multiple STF_INST_MEM_CONTENT records, where the first STF_INST_MEM_CONTENT record refers to the address specified in the STF_INST_MEM_ACCESS record, and subsequent STF_INST_MEM_CONTENT records refer to subsequent target addresses appropriately incremented to preserve continuity of data.

Trying this however, I triggered an assert in stf_lib, stf-inc/stf_inst_reader.hpp: 147

stf_lib/stf-inc/stf_inst_reader.hpp

Line 147 in eb394e5

stf_assert(desc != IntDescriptor::STF_INST_MEM_CONTENT,

                    stf_assert(desc != IntDescriptor::STF_INST_MEM_CONTENT,
                               "Saw MemContentRecord without accompanying MemAccessRecord");

It seems like stf_lib currently only supports a single STF_INST_MEM_CONTENT record following a STF_INST_MEM_ACCESS.

stf_lib was happy if I break larger accesses into multiple smaller (64bit or less) accesses with STF_INST_MEM_ACCESS and STF_INST_MEM_CONTENT pairs, but this doesn't correctly reflect the access being performed.