8b59cfa1437c: summarizing at orthogonal sections (appearing after F*2^-5)
  results: depending on w, but 3016dc9b6147 is a factor of 1.5 more accurate!

317e661464ee: without summarization
  results: the number of aforms just explodes

3016dc9b6147: summarizing at orthogonal sections (appearing after F*2^-2)
  results: slightly better accuracy than dd990fffecc1, also integrated R120 with size of tucker

dd990fffecc1: summarizing at orthogonal sections (appearing after F*1)
  results: integrated only easy orbits at the border, roughly the size of tucker

b89c8c1a007b: summarizing at orthogonal sections (appearing after F*2^-2, could make higher, because we summarize only when it is absolutely helpful…). conservative intersection (only orthogonal enclosures).
  results: slightly worse than 3016dc9b6147 on R350, slightly better on R1, instable rectangles explode before chance for summarization...

4001de64ff45: small stepsize (2^-11), intersection with pre-splitting
  results: R350: very precise (almost size of initial rect...)
           R1: very slim, but to the left out of the attractor
           R122: only setting where it seems to be able to flow away from the origin (also with smaller w-values)

40b881423c4a: small stepsize(2^-11), intersection with pre-splitting more conservative than 4001de64ff45 (for faster intersections), summarizing more easily compared to 4001de64ff45, summarize all sets while going up, try to catch more nontransfersal sets
  results: R350: with w=3 almost as precise as 4001de64ff45, R1: same x1, but larger x2 extent with w=3
           R122: does not seem to flow away from the origin...

89c50408fcea: first tries for automatic iteration, relatively accurate, stopped because of missing symmetry

1dcc81d900ef: improved over 89c50408fcea, but very slow due to (too?) small initial rects

d09d6462e646: rougher initial and flowing rectangles than 1dcc81d900ef

eadd79e80011: like d09d6462e646, but starting from critical rects + no sections below 8

50ad88de9631: p=5, w=5, should be accurate, mandatory sections below 8 6 4 2, started with 37ff2debf415, but with some older compiled code (eadd79e80011 probably) (real mandatory sections to avoid infinite loops when restarting before section) fixes for mandatory sctns in 2fca57d50e88 and b5acfabcb0d7, fixes against infinite loop in 9f0a47302507, enforce absent intersection due to too small step sizes close to the origin in 5f67e8d5b232, + very different intersection for the difficult ones in d440823d9d82, fix + larger stepsize for difficult ones in d440823d9d82, fixed origin in fd21fc5e53f4, handle nonexistent strongest direction close to origin in ce13fe0a00a4, fixes in 934e61565bf6, fixed aboveness in baaa471c0e83, last experiment 06e87eb11241 lead to too broad returns (i.e. lower branch extends up to more than +5
