First of all: thank you for posting what may be the only unpopular opinion on this thread. Someone said something interesting!
I think it hasty to draw policy conclusions based on descriptive statistics, there are so many nuances to quantitative sociology that would require further investigation to tease out. In fact, there is probably a limit to how well we can understand family structure and its effects on childhood performance from a quantitative standpoint because it's based on observational instead of experimental conditions.
For example, to what degree are we conflating marriage with some hidden or common causes? Perhaps the kind of people who make good parents are also likely to get and stay married (i.e., the selection effect), as is evidenced by underperforming children in stable step-families. Perhaps the quality of parents matters more than the marriage itself. Also, there is certainly an interaction effect between poverty and single parenthood that researchers are still disentangling.
I would be inclined to think the quality of a marriage overrides it's presence, but the data I've seen just doesn't make it clear either way. Even after admitting there is a lot we don't know about family structure and childhood outcomes, trying to determine the best course of action for an individual family from aggregate data commits the ecological fallacy. We cannot restrict divorce options based on statistical averages -- not only is that bad for the individuals involved but it's not necessarily better for society.
Removing no fault divorce seems like a recipe for disaster to me because it requires proving fault. This will return to us to pre-1970 condition of women being trapped in abusive marriages unable to prove their way out of them. Historically, women were successful in proving drunkenness, failure to provide, and to some degree later on, cruelty. Adultery and abuse, especially emotional abuse, were extremely hard to prove.
To your point about women's earning and job prospects, in fact to this date parental resources drop significantly after a divorce and job prospects for mothers are significantly worse for fathers after a divorce (on average, see my point above about ecological fallacies). One of the major contributing factors to childhood performance gaps in single parent households is the rapid descent into poverty brought on by single motherhood post divorce. If your logic is based on fulfilling a contract for the sake of the children, then a reasonable extension of that logic is that we should have child support (in either direction) in cases were divorce (wether fault or not fault) occurs.
As for the school bussing program, it might be unavoidable for a period of time because it turns out that school desegregation is still a major issue at the heart of many public school problems in the United States. Integration is one of the best ways to improve academic performance, but we never really finished integrating schools (even in states that skipped the whole separate but equal thing) because lots of white/middle class parents (understandably but possibly incorrectly) take a NIMBY approach to integration. This American Life has a great two-part series up on school integration that is worth a listen.