fst lines read
"tag tec mode,fmt",
where
-	tag is the field number used in the index posting
-	tec is the indexing technique (see below)
-	mode is a Mmc formatting mode or, for tec 5..8, a prefixing format
-	fmt is some format expression, typically a single field like v24


*	formatting modes
indicated by Mmc, where m is a character indicating some translations,
and c is a character indicating the handling of case; default MPL

m
-	P proof mode: no changes applied
-	H heading mode: angle brackets are removed,
	^x is replaced as ';' for x=a, ',' for x=b..i, '.' for others
-	D data mode: like heading mode plus '.  ' after each field

c
-	L lower case: no changes applied
-	U upper case: (ASCII-)characters are converted to "uppercase" as listed
	in the file ISISUC.TAB (a textfile containing 256 replacement byte values
	like ... '096 065 066' ... in 16 lines of 16 numbers each)


*	indexing techniques
-	0 one entry per line (format should produce appropriate newlines with /)
-	1 one entry per line or subfield (use proof mode MPx to retain ^x)
-	2 one entry per each term in <angle brackets> (use proof mode)
-	3 one entry per each term /between slashes/
-	4 one term per "word":
	a word is a sequence of 'alphabetical characters' as listed in the file
	ISISAC.TAB (a textfile containing numbers like ... '098 099 100 101' ...).
	"stopwords" (listed in the db's .stw-file) are ignored
-	5..8 like 1..4, but instead of applying a mode Mmc,
	every entry with delimiter 'd' is given a prefix 'p' like
	1 8 '/TI=/',v24 (prefix //-entries with TI=)