This PR does a few things that make project setup a little nicer (especially with nix, but also without nix) --

1. It includes dataframe-hasktorch and dataframe-persistent in cabal.project, which makes it possible to run cabal repl dataframe-hasktorch (or -persistent) from the repo root and drop into an appropriate dev shell
2. It includes dataframe-hasktorch and dataframe-persistent in the packages that the nix dev shell depends on, so e.g. ghci, then import Torch / zeros' ([3, 4] :: [Int]) works (i.e., ghci in the nix dev shell has all the dependencies for each of dataframe (not new), dataframe-fastcsv (not new), dataframe-persistent (new!), and dataframe-hasktorch (new!))
3. It fixes a boneheaded mistake I made earlier today where I tried to include dataframe-fastcsv from the dataframe-persistent.cabal file 🤦🏻 I'm not sure why the dev shell didn't throw a tantrum

Testing instructions

• dev shell improvements
  • nix develop --command ghci
  • do any hasktorch examples
  • do any persistent examples, e.g. copy the synopsis from the persistent book into persistent-example.hs, :l persistent-example.hs from ghci, then run main
• repl fixes
  • cabal repl dataframe-hasktorch, do Torch examples
  • cabal repl dataframe-persistent-tests (because of persistent-sqlite), then do the same persistent example from above, but :l ../persistent-example.hs from wherever it is relative to the dataframe-persistent directory