Since you feel more confortable testing the numbers for yourself, here's how
I set about disproving the jig Vs mana regen debate. Your comments are
welcome.

Set up:

L21 bard, L35 shaman.

Control:
Shaman burns all mana down until he cannot cast sense magic (5 mana).
Shaman stands, repeatedly attempting to cast companion spirit (350 mana)
until he doesn't receive the Insuffient Mana message. We can assume that at
the beginning of the timed interval the shaman had less that 5 mana and at
the end of the elasped time the shaman had greater than 350 mana. The
potential error variance between assumed "0" mana and casting mana
is -1.43%.

Test:
Shaman repeats the above, however the bard should commence playing jig prior
to the timed interval.

Results:
Times were near equal (delta: < 7 sec) however difference in regen rates
amoung classes/races/int/wis do not allow us to to draw a universal mana
regen rate.