CommAlgebras as CommRingHoms

[felixwellen]

This is an experiment to see if algebras work better when they are defined as homomorphisms of rings.
UPDATE: The experiment worked. It is however the case, that it is quite some work to port everything to the new definition of CommAlgebra and not all of this work is done in this PR. I think it should be merged anyway, since it is already very big and it should already be useful to others.

[felixwellen]

I did some random benchmarking (this PR is not really about type checking speed, but I made some changes while I rewrote code). In summary, the two modules in CommAlgebra with comparable content have signficantly lower type checking times now, while there is no significant change for the modules in Ring and CommRing I checked.

* Ring/Quotient
*New*

Deserialization            2,471ms (3,109ms)
Typing                         3ms   (580ms)

*master*

Deserialization            2,449ms (3,086ms)
Typing                         5ms   (535ms)


* CommRing/Localisation/PullbackSquare
*New*

Typing                                       1ms (8,193ms)
Deserialization                          4,001ms (5,061ms)

*master*
Typing                                       1ms (7,534ms)
Deserialization                          3,809ms (4,842ms)

* CommAlgebra/Localisation
*New*

Deserialization                         4,537ms (5,754ms)
Typing                                      4ms (1,357ms)

*master*
Deserialization                         4,721ms (5,998ms)
Typing                                      5ms (3,354ms)

* CommAlgebra/QuotientAlgebra
*New* 
Deserialization            3,431ms (4,312ms)
Typing                         4ms   (656ms)

*master*
Deserialization                         3,610ms (4,553ms)
Typing                                      1ms (2,131ms)

[felixwellen]

I explained this PR to Evan, whom I didn't ask to read through all of it. He believes me that it makes sense and I will just merge it now.