The new method uses a table and the taylor series for atan to compute
atan for real args.

qd-const.lisp:
o Add the two tables used for the atan computation.
o Include clisp code that generates the tables.

qd-fun.lisp:
o Implement atan using a set of tables and a Taylor series.

timing.lisp:
o Add a simple timing program to evaluate how fast each atan method
  is.

o Remove old version
o Update new version to skip reduction if the arg <= pi/4.

o Implement better pi reduction and also always do pi reduction.
o Add test for sin(pi) and cos(2^120), to test pi reduction.
o Accuracy for erfc is only 198 bits.

 qd-gamma.lisp::
 Add continued fractions

 rt-tests.lisp::
 Add a test for erfc

 * Simplify {{{sum-an}}} so we stop the sum when the terms no longer
   contribute to the sum.
 * Change {{{big-n}}}.  This still needs work.

   converge when v = 0.
 * Clear hash tables in bessel-j.

caused oct.elliptic-pi.n2.d to fail.

 * qd-gamma.lisp::
   * Return sqrt(least-positive-normalized-double) instead of
     least-positive-normalized-double.
 * rt-tests.lisp::
   * Add test for this case.

case where v is an integer.

 * Add series for Bessel J.  (Not working yet.)

series.

qd-methods.lisp:
* Define FLOATP
* Fix bugs in FLOAT:
  * (FLOAT float nil) is an error
  * (FLOAT float) returns the float
  * (FLOAT rational) returns a single-float.

qd-package.lisp:
o Export FLOATP, shadowing CL:FLOAT.

rt-tests.lisp:
o Add a few tests for FLOAT.

qd-methods.lisp:
o Define the macro

qd-gamma.lisp:
o Use it.