It would be hard. Not impossible. But you're not doing it in a weekend or a month. If you were to get something like this done, assuming you have a job or go to school, in a year, I'd be very impressed.
So most RPGs you level up through grinding. Kill this slime 100 times. Kill this hydra 14 times. Etc. But what if instead of the normal grinding, you leveled up through events. For instance, if you're in a cave and solve a puzzle, you gain Intelligence points. You move rocks around, enter Tournaments, and wrestle bears, you gain Strength points. You catch fish with your hands, you get agility. So on and so forth.
Would this be too tedious? Or an excellent opportunity for more [optional] quests?
should i have two types of quests: epic and mundane. mundane would not have the epic encounters, but would also not get the epic treasure. epic quests would be rare encounters. mundane tasks/jobs would be common quest encounters. this would allow the player to both take on mundane jobs for "pay" (fedex, escort, gather, etc), and take on epic quests for great rewards.
This is best course, IMO. Personally, I like to gain a little coin escorting Granny Babaluu to the grocery store without having to fight for my life in some epic or near-death experience. Some downtime quest-wise is a good thing.
Another idea is to have mundane quests lead to epic quests. For instance, you get a hint about an epic quest everytime you take a mundane quest. Once you have enough clues, you can go to where the epic quest is or the epic quest can be unlocked for you.
As an example to Spiro's point, make a simple calculator. One text-based, another graphical. It should do add, subtract, multiply, and divide. After you finish that, you'll be ready for your next project. Which should be creating an AI agent that uses pathfinding to find a random roaming bunny.**
So my question is, should I go ahead with Python, or teach myself GML?
You should do both. Make your game in GML. Take notes of what you're doing. Then after your Python class (or 2), try making the same game in Python. As @frob mentioned, as a programmer you'll be expected to learn and know multiple languages. Programming is never either-or; it's a toolbox of opportunity. There's more than one way to fix a problem; there's more than one way to build something.
From what I understand the Raspberry PI is close to what the older consoles were hardware and development wise. Why not make a game for that first before tackling the other consoles? Because, from everything that I've read, you need to know an extraordinary amount of assembly and computer engineering to make heads or tails of those consoles and their tech manuals.