This was my first concern. Some back-of-the-envelope calculations:

Photosynthesis is pretty lousy in terms of thermodynamic efficiency. About 1% of the light that hits a plant is converted to useful chemical energy. The plant will have to use most of that energy for its metabolic processes. Luciferase itself is a very efficient enzyme, however, so I'll generously assume that 10% of the energy that the plant captures can be turned into useful light. So the overall efficiency can't be much higher than 0.1%. By comparison, solar cells are around 10% efficient, and LEDs 20%, so at first glance the luciferase plant seems to be an order of magnitude less efficient than the solar powered flashlight my in-laws gave me for christmas.

In absolute terms, there is about 100 watts/meter^2 of energy in sunlight. If you've got a one-square-meter window full of the hypothetical plants sitting in sunshine all day, let's say they can absorb 1500 watt-hours, and then convert 1.5 watt-hours into useful light. That'd be comparable to running a 5-watt LED for an hourish, which could be useful if you could turn the luminescence on and off at will. But if the plant is glowing all night and only a portion of the light is emitted in a useful direction, maybe the window-full-of-plants would give off light comparable to the little cluster of LEDs on the front of my computer. So overall I'd say that the idea is not completely impossible, but still totally impractical.

A "four year thesis project" sounds just about right to me. My particular program in biology has an average completion time of 5.5 years, and the national average I believe is 6.5 years. But I don't even have a formal thesis project until I write a proposal for the end of my second year. I spent my first year taking classes and doing research rotations. I'm currently in my second year, I've picked a lab and started to do some productive research, but I still have classes and TA duties, and I spent a lot of time this year learning the techniques that I will need for my thesis. So yeah, I'm aiming for a "four year thesis project".

I'm a little shocked that 10 or 11 years is at all normal. In my experience, anything beyond 7 years is the stuff of horror stories. 10+ years must be in fields where a grad student has to do their research after working as a TA for 30+ hours per week to afford their ramen.

Have you ever actually used a microscope? You can't just put one on an arm and wave it over a patch of dirt and expect to see anything interesting. Microscopes capable of resolving bacteria have a very tiny depth of field, so if you point it at a patch of ground there'd be nothing in focus. You have to take samples, mount them on a slide, stain them (usually), and then place them under a microscope. And on Mars, most potentially interesting microbes will probably be buried. To do that you have to have some fairly sophisticated sample handling mechanisms... which is exactly what this new mass spec instrument doesn't need. Now you just point it at an interesting patch of dirt, zap it with the laser, and suck all of the vaporized ions into the mass spec.

Part of the problem in my experience is that there isn't a single, reliable cloning manual out there that provides start-to-finish logical guidance. Most people do their projects in an ad-hoc sort of way. If I have a particular cloning project, my resources are (in order) my PI, my labmates, the NEB catalog, and whatever I can google up.

You'll probably want a decent background in molecular biology -- equivalent to one or two intermediate college courses, or whatever you can self-learn from the right textbooks. Molecular Biology of the Cell is one of the better textbooks out there. There are some relevant manuals on cloning out there, but mostly they're concerned with the gory details on what biologists do at their bench. One is written by Maniatis. It doesn't really present a unified, logical approach, but is really just a collection of protocols and recipes for accomplishing individual steps, without providing much guidance for a whole project. Still, if you have access to a university library, it's probably in the reference section, and it has a lot of good information covering the basic theory of all sorts of techniques. Another good resource is the New England Biolabs website -- they sell reagents and kits for subcloning, and their technical references are excellent. So good, in fact, that their catalog was passed around one of my undergrad lab classes as a supplemental textbook of sorts.

The big challenge for you is to put all of that crap together -- distilling the accumulated lab superstition and hodge-podge of tools into a flowchart of decisions that outputs useful cloning advice.

Unfortunately, most such attempts are written by biologists with a bit of coding background... and end up being the crappy sorts of projects that I could do myself, if I wanted to put the time in. The better computer scientists making their way into biology are all going into bioinformatics, which involves big sexy problems like analyzing and comparing whole genomes. The routine concerns of everyday biologists haven't attracted the right talent.

Back to school time for me means I have to elbow my way through hordes of clueless freshman on the way to my building, and I have to put up with smaller hordes of clueless first year grad students coming through my department. Luckily I live off the beaten track, so I only have the battle the usual hordes along my commute. Thankfully for now, that's the extent of my contact with the "school" part of academia.