Watch Table get slower with more recent commits.
Our command is always: Table(a(i),[i,10^5])

free symbol list is present the whole time, so thats
not the culprit. Maybe how its used.

commit                                     Time
486a76c0ca0f6b04f5cc05589eff0587ee5a2f6a   1.81 s
64e597c5eeac53264110dd72a217d5edf4e685f7   1.43 s
7d1b9e4aa70c6c5adf89c9991c933d7f5ca48c9a   0.75 s
83cd8411472d83d86c4a932f0b1c363d03328e6f   0.68 s
1ac6fbe717c56cf8338ab70d809a2360ca60b2c1   0.63 s
7043cf1101f2fe48a548639693be1f7165ea9a58   0.53 s

# This commit has a comment about Table running 20% faster
80e09ff0b6869af90107c3dc359c46e78464d86f   0.55


# I try disabling many evaluation steps (essential ones).
# Nothing lets the speed approach the old speed.


# Table present in this commit, but not TimeOne or Timing
# It runs pretty slowly, more than a second
commit 7fd3ea5da62a7b32f6013e9cb8bc2dd9573bcc0e
Date:   Tue Feb 3 21:17:36 2015 +0100

commit f3b5cacdb3bee7d865ff7b90c1b87e1f1c7e613f
Date:   Sun Feb 8 12:07:08 2015 +0100
sjulia> Table(a(i),[i,100000]);
elapsed time: 0.631463795 seconds (115963592 bytes allocated, 13.01% gc time)
But, we are cheating here and using meval, not doeval.
