Since it is stated that the extra current persists for many minutes, until the breaker trips, I rather doubt that a defective ballast is to blame.
The extra current, beyond the expected load is about 10 amps, that is well over a kilowatt, if that much extra power was being disipated in a ballast, it would have gone up in smoke long ago.
I would stronlgy suspect a defective underground joint allowing earth leakage, this may get hot locally, but would be concealed by normal temperature soil above, and thus not be readily located.
Such faults can be challenging to locate.
One possibilty is a voltmeter and a long lead.
Connect one side of the voltmeter to a good ground, such as the service entrance ground bar. Connect the other to a metal spike which is inserted into the earth at different points alonlg the cable run, the fault is probably close to the point that gives the highest reading.
This is by no means completly reliable as leakage/ground currents from elswhere can give misleading results, and of course the circuit must be energised for testing.
It would be advisable to wear linesmans gloves when using the tempory test connection, in case the faulty cable is very close, and a dangerous voltage is present.