It didn't work for values over 0xff. Duh!

It doesn't serve much purpose and takes up many precious bytes. And,
after all, parsed didn't do it either...

2DROP was left in boot.z80 because it's used in icore.

This allows us to remove forth/z80a.fs. Another big step towards
self-hosting!

It seems that I had forgotten in an earlier commit.

This has been tested in the RC2014 emulator. I could edit blocks with
the Block Editor and they would correctly be written to SDC.

Goes through initialization in RC2014 emulator.

As I wrote in my "Clarify signed-ness" commit, "0 <" is broken.

Also, made this unit a bit more compact. The RC2014 stage1 can
really use some breathing room...

All of that should help me implement sdc support much faster in
Forth.

I now get to "fls" in recipes/rc2014/sdcard!

so far, it goes through initialization with the rc2014/sdcard recipe.

Also, fix some PSP leaks related to LOAD.

as well as emulator support for it. We can now write an edited
block back to "blkfs".

Saves quite a few bytes in the final binary.

It's more hassle than it seems to move the stable ABI...

I forgot to do it earlier. NOP having an opcode of 0, it didn't have
any effect.

Should be significantly faster now. Less push/popping and usage of
DE instead of IY.

Saves quite a bit of binary space. More than 300 bytes in forth1.bin

Remove flagsToBC boot routine.

Makes boot binary a bit bigger, but that "_pdacc" word will be
reused in high level apps.

It's not used

The former was only used in the peculiar context of "/emul". The
regular case is actually HERE pointing to RAMEND on boot.

This allows optional subsystems to not needlessly reserve too much
system memory and it also reduces offsets bookkeeping.

Let's revisit the relinking approach, with LITA this time.
This reverts commit 6652125d47.

It seems that xcomp can't manage to reliably compile core+ units.
Let's limit the role of xcomp to z80c+icore.

I'm getting a little desperate here. I'll revisit the link.fs approach
with a new idea I had...

Replace its usages with direct RAM+ offsets. The (sysv) mechanism was
incompatible with cross-compilation of a full interpreter.

It's not needed anymore now that the boot binary is built in one pass.