new syntax for exact mp floating constants
     decimal_int ( "p" | "P" ) [ "+" | "-" ] decimal_int
     the value is decimal1 * 2^decimal2

R=r
OCL=20357
CL=20357