The same question could well be asked of the LM's descent engine and the main engine on the Apollo service module, however, which did both need to fire in free-fall. In those cases, the smaller RCS thrusters on the LM or CSM were fired first, to "settle" the tankage and separate the fuel from the helium. In the LM case, this "ullage burn" was about 7.5 seconds. The first couple of service module burns -- typically for mid-course correction while en route to the moon -- generally didn't need an ullage burn prior, as the tanks would be full of propellant with little or no volume of helium. SPS burns later in the mission did require ullage burns. The RCS thrusters produced about 100 lbs of thrust each, and four would be used for the ullage burn, yielding roughly 1/200 g acceleration.

The same RCS ullage burn technique would also apply to a situation where the descent engine failed and the ascent engine needed to be used for abort from free-fall, or in flight testing of the ascent engine.

That, in turn, raises the question of how helium ingestion was avoided in the RCS thrusters, since they were also helium pressurized. In those cases, the helium was separated from the propellants by a teflon bladder, so the helium didn't mix with the propellants. This was more practical to do on the smaller scale of the RCS propellant tanks than it would have been for the larger engines.