Fix three citation keys that do not resolve

[Chessing234]

Same class of issue as #141. Each of these keys appears in a page's bound tables or comments but matches no entry in that page's own reference list, so the citation dead-ends for a reader.

page	cited	defined	resolution
22b	[O2013] (8.50 upper bound)	[Olsen2013]	Olsen & Bohr, A principle for ideal torus knots — the double-helix construction the row's comment names
46a	[Ta2004] (×2, in comments)	[Tao2004]	Tao, Some recent progress on the restriction conjecture — the survey both sentences point at. [Tao2003], the bilinear estimate paper, is a separate entry already cited correctly elsewhere
52a	[AS2000] (3.26 lower bound)	[AC2000]	same Achlioptas–Sorkin paper

For 52a I changed the reference entry rather than the citation: [AC2000] is cited nowhere on the page, and the authors are Achlioptas and Sorkin, so AS is the key that matches the page's own convention (compare [KKSVZ2007] for Kaporis–Kirousis–Stamatiou–Vamvakari–Zito). The entry also spelled the first author "Achioptas"; that is corrected in the same line.

How I found it: scripted every page's cited keys against the keys its ## References section defines, over all 111 constant pages. After this change exactly one page still reports:

• 56a uses eight keys ([BB2011], [JS1981], [KiSh2002], [KS2003], [Li2006], [LRS1999], [MS2004], [Sar2005]) and has no ## References section at all. I have deliberately not touched that one — supplying those eight entries means identifying the papers, and guessing at citations in a reference database seemed worse than leaving it visible. Happy to open it as an issue, or to add them if someone can confirm the intended sources.

One concern only; no bound values changed.

[Chessing234]

closing this myself — on reflection it's reference bookkeeping, not worth your review time. the 56a page having no references section at all is the only part that actually matters and i'd rather raise that properly than bundle it with key renames.