The case was on shaky legs from the start. A circumstantially based prosecution better have a butt load of circumstantial evidence that excludes any other possibility. If it doesn't, it's going to get picked apart and render itself to alternative theories to a jury. This prosecution had no direct evidence. It had a lot of very disturbing trends though, but nothing that could directly prove she killed her child. That's reasonable doubt, and even more so when you start throwing experimental testing and evidence into the mix.
The testing of the hair and the testing of the trunk for evidence of decomp are FAR from tried and tested criminalistics. They are also far from tested legally either until these procedures can pass a Daubert or Frye test. I'm kinda shocked they even made it into court without a Daubert challenge. I suspect based on the description of the hair and decomp tests and what they can render as evidence, they wouldn't have been considered direct evidence anyway, so the defense didn't Daubert challenge them.
I don't think based on the publicity of the missing girl and the investigation, the prosecution could have continued to postpone the prosecution. They just didn't have enough the evidence to close the deal. I suspect she either did the killing or was involved in the killing before and after the act based on the circumstantial evidence, but I couldn't have committed to a guilty verdict if on the jury either. Their was way too much drama and way too substance.

