Make great documentation goal 1.2, and renumber goals 13 to 16 as 12 to 15 wherever they are cited

docs/decisions.md

@@ -171,7 +171,7 @@ so no path repeats it. Valid until a locale's lowered name clashes with another'
 
 2026-10-03, a restructure the maintainer approved; widened to any source on 2026-10-09.
 
-Goal 13: `New` pays nothing for a category a run never reaches. A manifest,
+Goal 12: `New` pays nothing for a category a run never reaches. A manifest,
 `.fejkdata.json` at a source's root, may carry an index: each category's parent table and
 the paths `List` advertises below it. A category of an indexed source loads with every
 category it references and its whole table family, and binds as a whole load does, so
@@ -446,7 +446,7 @@ cost.
 
 ## The performance gate asserts allocations, not wall-clock time
 
-2026-09-03, Lilleman auf Larv. Goal 13.
+2026-09-03, Lilleman auf Larv. Goal 12.
 
 `AllocsPerRun` is deterministic across machines, so a 10% ceiling does not flake under
 CI load, while time varies with the machine and its neighbours. A rendering slowdown
@@ -459,7 +459,7 @@ heap through `pickKey.under`, holds at its baseline with no margin: from a basel
 
 ## Rows live in a TSV, the shape in JSON
 
-2026-09-17, Lilleman auf Larv. Goals 8 and 13.
+2026-09-17, Lilleman auf Larv. Goals 8 and 12.
 
 `New` allocates once per node, so a register of thirty thousand rows written as JSON
 objects would cost it a second; a TSV is one allocation whose cells are substrings, and
@@ -579,7 +579,7 @@ generators of their own.
 
 2026-09-17, Lilleman auf Larv.
 
-Goals 10 and 13: a path that fails moves no seeded stream, at the cost of one draw-free
+Goals 10 and 12: a path that fails moves no seeded stream, at the cost of one draw-free
 walk per call, compiling its steps into a stack buffer of 16. Valid while that walk
 costs little against the draw it guards.
 
@@ -607,7 +607,7 @@ them; a data package imports only the packages it reads.
 ## The default embed holds every Swedish postort the import can place, with all its streets, and, of the US places that hold most of a postal code's land, every state capital, every place of 100,000 or more and a population-weighted tenth, each with a quarter of its streets, at least ten
 
 2026-10-11, Lilleman auf Larv; revised the same day, after a tenth of the streets gave
-Göteborg 163 and Stockholm 37. Goals 2 and 14: ten streets per place put 500 Stockholm
+Göteborg 163 and Stockholm 37. Goals 2 and 13: ten streets per place put 500 Stockholm
 customers on ten streets, and US places of 25,000 or more left out every town and
 village. `geo-us.py` draws with Efraimidis and Spirakis's weighted sampling. It draws
 only among places holding most of some ZCTA's land, since no other place can carry a
@@ -620,7 +620,7 @@ else by a name no other place in its state carries. No census-designated place s
 so Honolulu is the one capital missing. Every Swedish street is 174k rows; a quarter
 of the US streets 475k. On 2026-10-11 the embed grew from 5.2 MB to 16 MB, and a CLI
 run printing the first `en_US.address` took 145 ms, `sv_SE.address` 95 ms and
-`misc.territory` 3 ms, within goal 13. `--place-share`, `--street-share` and
+`misc.territory` 3 ms, within goal 12. `--place-share`, `--street-share` and
 `--min-streets` on the import scripts build another set. Valid while the first reach of
 a country stays unnoticed beside a test run.
 
@@ -748,7 +748,7 @@ frequency it is read, as `misc.timezone` reads GeoNames populations and
 2026-09-24, Lilleman auf Larv.
 
 The weight is the population GeoNames records in the zone's cities of 15,000 or more,
-floored at 15,000, which goal 16 asks for: over 300 seeded draws of
+floored at 15,000, which goal 15 asks for: over 300 seeded draws of
 `misc.territory[US].timezone`, the four zones most Americans live in took 278 where an
 even weight gave them 44, and `America/Indiana/Petersburg`, a town of 2,400, fell from 17
 to 0. Valid while a draw weighted this way lands where people live.
@@ -855,7 +855,7 @@ the direct chain is the one a reader can predict from the tables' parents.
 
 2026-09-30, larv-review on the comprehension round the maintainer approved on 2026-09-29;
 replaces "The path walks are separate loops".
-Goals 3 and 13: `drawSteps` draws every path from its steps, compiled when templates resolve, by
+Goals 3 and 12: `drawSteps` draws every path from its steps, compiled when templates resolve, by
 `compiledPath` or, for a caller's path, by `callerPathSteps` into a stack buffer. Both take each
 step through `compileStep`, and differ only at a choice: `compiledPath` walks every variant,
 `callerPathSteps` the first. On 2026-09-29 one loop over all three walks, switching on a mode field,