* parser: fix handling of FLOAT(p) and FLOAT(m,d) types * FLOAT(50,4) should not automatically become a DOUBLE, it is just a FLOAT shown with 50 digits. * FLOAT(0) and FLOAT(24) are the alias of the same type FLOAT. There is no need to record the Flen. * types: fix Restore of DECIMAL(m,0) types * tests: add test cases