Well... let's start with some incredible news (
): orbitals don't exist.
They're simply the area of the space in which the probability (given by the square value of $$ \psi /$$, which is the wave function of the electron - I'm referring to Schrödinger's equation - ) is bigger than 90%. Physics consider it quite a picturesque fact that chemists call these areas "orbitals" as they've got no physical reality and moreover they're not even linked with the beahviour of orbits. Heisenberg's principle tells us that we can't really know where the electron is and how fast is he moving (nay: that the indeterminacy of the first measure is inversely proportional to the indeterminacy of the second one, so that if we increase precision in one the other is less accurate) so all we get is a probability of presence, it's more likely to find electrons there than anywhere else. Here is a visual aid to make you understand that there's no fixed area (no dumb-bell) but only a distribution:
1) So answering your first question the number of electrons doesn't affect the shape of obitals and we could also say that orbitals as we know "exist" even if there are no electrons on that level (every atom has a number of orbitals, obviously we don't bother to think what would the behaviour of our electron be if there was an electron on level 4f in H)
2) Yeah, because naturally unpaired electrons make atoms less stable, this means more potential energy and a bonds decreases it. As in nature a system always tends to the less energetic configuration it's easy to understand that a bond will soon be formed with another atom or molecule. Just think at the high reactivity of the radicals (an easy example of unpaired electrons).
3) You don't know that there is hybridisation as there's no real hybridization in a bond. It's a way we use to think at the bonds and understand them according to the VB theory. If you use the MO theory you won't find any sp
3 but bonding and anti-bonding orbitals. So I think we could say that hybridisation accords quite well to experimental evidence especially when you have to easily plot a molecule's geometry but it's not based "purely" on that. Most of the molecules have resonance forms and the hybridised one is just one of the possible; while in other cases you can only use hybridisation to justify the bonding of an atom with other atoms (think at Xe compounds).
I'll leave the other three to someone other or I'll post again later... see if these answers are clear of if you nedd some other clarification
Edit: I see Yggdrasill it's a turbo-poster
Well, I hope my information are useful anyway...