Cellular automata don't use exclusively the rules 90 or 150. This classification is due to Wolfram, cf.,  also, http://www.stephenwolfram.com/publications/academic/algebraic-properties-cellular-automata.pdf

For a review of computer simulation techniques, cf. here: http://iopscience.iop.org/0305-4470/24/5/007/pdf/0305-4470_24_5_007.pdf

The number of rules depends on the number of neighboring sites-it's just counting boolean functions. So, if a rule is of the form s_i^(t+1)=f(s_(i-1)^t,s_i^t,s_(i+1)^t), it's a boolean function of three boolean variables. There are 8 possible input configurations and 2 possible output configurations. So the number of possible rules in all is 2^8=256.  

This means that it's possible to associate to any such rule  an integer in the range 0 to 255. If one writes this integer in base 2, one can deduce the rule. For instance, rule 90 means

90 = 64 + 16 + 8 + 2 = 2^6 + 2^4 + 2^3 + 2^1

The exponent encodes the values of the string (s_(i-1),s_i,s_(i+1)) that lead to output 1 for s_i^(t+1). 

6 = 110

4 = 100

3 = 011

1 = 001

which can be recognized as the XOR function.  

The corresponding calculation leads to the function for rule 150. 

The study of the correlation functions then leads to the dynamics and addresses the question of what the output is like-point attractor, periodic structure, strange attractor. The two typical quantities studied are the ``magnetization'', or ``activity'', i.e. the average value of the  number of sites that have the value 1 (or the value 0), suitably normalized; and the ``spreading of damage'', i.e. the behavior of the Hamming distance between two configurations: does it go to zero, does it stay constant, and so on. 

In the link below you have a very nice introduction to Cellular Automata, including the rulesets for Elementary Cellular Automata with one dimension neighborhood that leads to 256 rules.

http://natureofcode.com/book/chapter-7-cellular-automata/

Best regards.

Only that, unfortunately, it doesn't implement multispin coding (cf. Stauffer's article linked above). 

Thank you Nicolas and Mario for passing information.....

Cellular automata don't use exclusively the rules 90 or 150. This classification is due to Wolfram, cf.,

According to Wolfram CA, there are 256 rules..

A cellular automata with 'K' colors and radius = 'r', there exists K^K^(2r+1) rules.

For example, in the case of 3 neighborhood binary cellular automata ie, k = 2, r=1,

there are 256 rules exists, among them K^(2r+1) ie 8 rules have additive properties, they are , rule : 0, 60, 90, 102, 150, 170, 204, 240 , choosing any of these rules for generating hybrid cellular automata will give 'maximal length cellular automata'.  The neighborhood dependencies  of rule 90 and 150 differ only in one position, so its a better option than other additive rules.

As far as your second question is concerned: there are countless CA rules. If you think about it, you can have countless rules for any number of states, any kind of neighborhood and any dimensions. 

Try to think of an n-dimensional CA, given a very large n. Toffoli showed in 1977 that any CA of dimension n can be simulated by a CA of dimension n+1.