Unless you have a solid project in mind, buy a board that has the features you initially want to play with, and focus on learning the tech with minimal distraction. Maybe the best thing would be to find the material that suits your learning style, and then get the board that works with that material.
It is very unlikely that your first board will be ideally suited to a big project, unless you have a really good idea of what you want to implement or the board is over-speced for what you need. Seriously consider getting a very low end board to get started with, one which is USB powered makes it as simple as plugging in a cable.
Just don't get a Digilent Basys2, the board is very old, the FPGA small I think it is only in production because of the course material based around it. It also has a very poor oscillator on it. The Nexys range are great.
Perhaps one of the biggest disappointments for low end boards is the FPGA <-> Host bandwidth. If you require more than fast RS232 can supply then choose carefully!
It usually turns out that whichever toolset you learn with you will most likely stick with. I learnt on Xilinx ISE and find that better than Altera's Quartus or Xilinx's Vivado, but then Quartus fans feel the opposite. It's a bit like programming IDEs, by the time you have played with three or four you will start to see them as being pretty much equivalent and you focus on tasks not the IDE.
I really like my new Basys3, even if it forces me to use Vivado tools. The Artix7 FPGA is fast, relatively large (comparable to the Altera one ona the DE-0 Nano), and the board has lots of simple I/O to play with. The DE-0 has DIP switches and a couple of "must use a biro" push-buttons - enough for config settings for a CPU but not that easy to play with.
Oh, and trust me, as a beginner you will not get SDRAM working outside of a SoC design - you will need a lot of study, experience and troubleshooting skills to get that far.