Overall conclusion
In Opera it is possible to process all test cases, however it is not possible to create a single algorithm. Opera has the best results of all the browsers, it all goes down hill from here. Mozilla Firefox starts off well with patterns matching those from Opera, however half way through they diverge and go down their own path (and not a good one). It was hard enough to analyse the results for these; it's actually impossible to process them. Internet Explorer has a fatal flaw in that it separates cells with spaces rather than tabs. This makes it impossible to process anything from Internet Explorer, however if it simply replaced the basic building blocks it uses to those Opera uses, the results would be identical to Opera's! Safari is where I really started loosing hope, only a small portion of the results can be processed, for the majority of them it's just not possible. Konqueror's results were just simply terrible; it's not possible to process any of the results for it at all.
Ultimately there's not a single browser that provides me with a good enough set of results. Opera is the closest, but it's not quite there.
The remedy
Opera is the closest to giving me what I need. It is possible to process all patterns it produces, however, it is not possible to combine processing into one single algorithm, which is not good enough! Either Opera needs to simplify things even more, by making the <p> tag have no effect on table formatting at all (in the clipboard). Or, some other solution needs to be sought which remedies the situation. Then the other four browsers tested need to follow suit.
Time to file some bug reports!...