The answer is it depends. Fill factor of 100 is fine for a clustered index on an incrementing value if your adding data and not updating or deleting. A fill...

How much of the 32 GB is SQL Server using?

Any versions AFAIK

You can create the table in SQL Server before the import matching the data types and lengths from the source. The import wizard reads the first few hundred rows to determine...

Using the row_number() function in a view doesn't add a lot of overhead however adding an identity column to the target table might be faster.

I've used this method on tables with millions on rows, it runs pretty fast.

If P:\ is a mapped drive then it won't be available to SQL Server. You will need the UNC path.

Without you providing a create table statement and sample data in the form of inserts very generically:

select non_agregated_column, max(date_column) MaxDate 
from tbl
group by non_agregated_column;


With length delimited file I define one column with the total length in the openrowset, and use substring for individual columns.

So create logins for al the groups in the domain

Does sys.xp_logininfo return the FSPs?

It could be done that way. One could also use a stored procedure for inserts. Are you planning to keep an Access database as a front end?