Scramble banks from the end of the ROM

[Rangi42]

This is more likely to test edge cases, such as having content in banks with their highest bit set.

Fixes #1036.

#1149 proposes allocating sections from the end of their individual banks. This is a similar approach, allocating section banks from the end of their overall ROM space.

Resolved

[ISSOtm]

Looks good to me!

Not actually good yet

[Rangi42]

Okay, now trying the next bank is updated too, not just assigning the first bank. Say you pass -S romx=3: then it will try the ROMX banks in order 3, 2, 1, 4, 5, 6, 7, .... Or -S sram=4: they'll go 4, 3, 2, 1, 0, 5, 6, 7, .... (And without scrambling they'll just ascend starting from firstBank.)

Resolved

[ISSOtm]

Subject to testing by @eievui5 before release, preferably?

[Rangi42]

The macOS tests give this error:

./test.sh: line 14: rm -f ${otemp@Q} ${gbtemp@Q} ${gbtemp2@Q} ${outtemp@Q}: bad substitution

It's complaining about this line:

trap "rm -f ${otemp@Q} ${gbtemp@Q} ${gbtemp2@Q} ${outtemp@Q}" EXIT

I haven't changed the definitions of those temp files, so not sure what the actual problem is.

My hypothesis would be that this line is responsible, since the rest of the new test is straightforwardly like the others:

rom_size=$(wc -c <"$gbtemp")

Except that tryCmpRom does the same thing without issue.

Edit: Turns out macOS needed the printf after all; even doing wc -c <filename instead of wc -c filename, it doesn't just print the size in digits.

[Rangi42]

Subject to testing by @eievui5 before release, preferably?

Tested and approved!