You are using terms in a loose and nontraditional way.
The argument isn't logical so much as it is valid. It isn't sound, based on the untrue premises, but it preserves the truth of the premises.
http://en.wikipedia.org/wiki/Soundness
Occam's razor provides a reasonable metric for burden of proof. Any position that requires Saturday to have a special property that affects light switches is obviously more complicated than the null hypothesis that light switches are agnostic with respect to the day of the week. The valid arguments in favor of a special Saturday are not terribly satisfiable (in a technical sense), and are thus unfavored.
A series of statements that are valid are logical. The conclusions can be wrong, as "anything follows from a false premise" (about the only actual statement I recall from logic lectures; see eg http://www.dpmms.cam.ac.uk/~wtg10/implication.html).
Occam's Razor is unsound. The truth is not necessarily simple.
Not really. This, for example, is a perfectly logical argument:
It does not require either of the premises to be proven (or even true, or even semantically meaningful) to be rational.To be reasonable, on the other hand, is a much fuzzier concept.
You're invoking Sagan's dragon. The burden of proof is on you, not on reality. Until you prove it right, you are wrong and have no argument to make.
"Burden of proof" has no real connection to rationality as such (beyond determining whether the logic itself is valid). It is fundamental to the scientific method, but that's another thing altogether.