Add process_lhs() method for Transform

[Anssi Kääriäinen]

The Lookup class has two methods, process_rhs() and process_lhs(). Typical usage is:

def as_sql(self, compiler, connection):
    lhs_sql, lhs_params = self.process_lhs(compiler, connection)
    rhs_sql, rhs_params = self.process_rhs(compiler, connection)
    # return the sql

The typical usage for transforms is:

def as_sql(self, compiler, connection):
    lhs_sql, lhs_params = compiler.compile(self.lhs)
    # return the sql

I think it could make sense to add process_lhs() to transform base class. This way one could write transforms using the same process_lhs() method that is used for Lookups, too. At least for me it is hard to remember when to use process_lhs() and when to use compiler.compile.

[Josh Smeaton]

Could I suggest going the other way and standardising on compiler.compile rather than process_X? I'm not sure how much work process_lhs does in the normal lookups, so it may not be feasible. But I'd like to standardise on the expressions way to ease the gap between them.