fix(core): Move the zenkaku width (全角幅; zw) unit into core

[ctrlcctrlv]

Where it belongs.

This touches on #1245.

This is blocking #1444, somehow (the manual is extremely brittle as all know well):

[font-fallback] Glyph あ not found in Gentium Book Basic
! Some glyph(s) not available in any fallback font, run with '-d font-fallback' for more detail at documentation/c08-language.sil:244:1: in \subsection after in \book:subsectionfont

! Unable to measure character at:
	documentation/c08-language.sil:244:1: in \subsection after in \book:subsectionfont
	documentation/c08-language.sil: in <file> (\)
	documentation/c07-settings.sil:51:1: in \includetable: 0x40ddce50
	! debug.traceback() did not identify code location
make: *** [Makefile:1451: documentation/sile.pdf] Error 1
make: *** Deleting file 'documentation/sile.pdf'

After much ado I discovered that the problem is that the unit zw (which stands for the width of a zenkaku—i.e. 全角幅) was popping in and out of existence based on the language setting of the document. Of course this won't work.

All units must be permanent. The concept of a zenkaku is not even Japanese-only.

[alerque]

Thanks for taking a stab at this. My (lack of) expertise on CJK has definitely shown through in the hacks in the manual. Speaking of which there is at least one more than needs removing if this works right...

[ctrlcctrlv]

@alerque I still want to rewrite ruby.lua, it's a longterm goal of mine. My private version is pretty cool:

I call these bizarre rubies 重ねルビ (kasane-rubi) which means "stacked ruby", or 𝑛重ルビ (𝑛-stacked ruby).

But historically they are not rare.

法令全書　明治７年　第６３ページ

台湾語仮名

韓国語での曲の翻訳

But my private version fails at some very important cases. I think the recent work on math.lua, and patches by @Omikhleia, will help a lot. I wouldn't want to release a version that doesn't meet most needs.

[alerque]

Oops. My modifications would have been smart if SU.error() was raising an error. It is not. Furthermore it is failing to do what it was supposed to do (finish off a valid PDF file on an aborted run). I'm looking into that...