Improve precision and scale behavior across adapters

[MasterOdin]

PR is a continuation of the work done in #2377, bringing that change across providers as makes sense. As part of this, now:

1. For sqlite, we will apply (p,s) args if precision is not empty and scale is not null, so that doing (10,0) is possible (same as mysql)
2. For sqlserver, if precision or scale is defined and you're using a decimal, than we will apply the (p,s) args with an appropriate default of (18,0) (defaults used as defined in the sqlserver decimal docs) is one is omitted. So it's possible to do just 'scale' => 5, and then precision will be implicitly set to 18, and you'll get (18,5). This brings its behavior in line with how postgres works (though it has a default of 10 for precision).
3. For sqlserver, when fetching columns, for decimals, we set precision. We also now always set scale as well (though it's expected that this will only be defined for numeric types anyway).

As part of this, I did learn that under the hood, precision and limit are equivalent for defining a column, which does prevent just using a precision for mysql and sqlite. It would separate out the behavior, but this would be a breaking change and thought would need to be made on the best approach.