I was thinking of allocation to be handing over a chunk of raw memory and initialization to be things like setting ref count to 1, [...]

Then allocation and initialization are done by the same function (not opcode). newSV and newAV, for example.

Surely allocation of a new array would only be done on entry to the block at the start of each iteration?

No, it's done on exit.

The directive to which I referred is SAVEt_CLEARSV. It's placed on the stack by SAVECLEARSV. In the code relating to SAVEt_CLEARSV in Perl_leave_scope in scope.c, you'll spot

switch (SvTYPE(sv)) {    /* Console ourselves with a new value */
case SVt_PVAV:    *(SV**)ptr = (SV*)newAV();    break;
case SVt_PVHV:    *(SV**)ptr = (SV*)newHV();    break;
default:    *(SV**)ptr = newSV(0);        break;
}
[download]

If you would care to point me to documentation where I can read some more that would be helpful

I don't know where or if this detail is documented. I discovered it while searching for my's runtime effect.

perl -MO=Concise -e"code" gives the names of the opcodes (e.g padav for my @array), which you'll find in the pp* files.