Was this the aim for all APollo EVAs initially, or just Apollo 11? (I find it hard to imagine an H or J mission with just one person outside)
I don't think they were thinking that far ahead.
You do have to consider that communications between the two astronauts was more important than comm between either and the earth. That ruled out a LM relay between the two astronauts; they had to talk direct in case they were too far from the LM. By having the CDR relay the LMP to the LM, direct communications between the two could be maintained with just one VHF receiver on the LM.
Nowadays you'd use a mesh network where every station (LM, LRV, CDR, LMP) transmits and receives on a single channel and they automatically relay high speed digital packets in all directions as conditions permit. Or everyone would have two mesh radios on separate channels for redundancy.
When did the change from one person to two people on the lunar surface happen?
I know I've seen that discussed (that's how I know the reason for the relay) but I can't remember where.
Did post Apollo 11 EVAs also relay through the commander's PLSS?
Yes, all of them. The only change came with Apollo 15, when they had the relay package on the rover in addition to the LM. The two were almost functionally identical, including the ability of the rover relay to transmit either PM (for voice) or FM (for voice/video) S-band to earth. PM had much better link margin, just as it did from the LM, so it was used through a low gain antenna during drives. FM required the LRV to be parked and the high gain antenna manually pointed at the earth.
About the only difference is that the LRV PM transponder was not coherent, as it listened to the same uplink from earth as the LM but transmitted on its own downlink frequency. (The uplink and downlink frequencies in a coherent transponder have to be in a specified ratio of 221/240 for S-band). The ground could therefore select a return (astronaut->Houston) link by just switching earth receivers.
Selecting the forward link was a little more complex. Uplink voice was sent to the LM (and CSM) as NBFM on a 30 kHz subcarrier. A separate subcarrier of 124 kHz was used for the LRV, so the ground could select a relay by just modulating the appropriate uplink subcarrier. The LM and LRV transmitted to the astronauts on the same frequency, but voice-activated switching (VOX) keyed each transmitter only when the associated receiver was active; this avoided interference on the moon between the two VHF transmitters.
Commands were sent to both the LM and LRV on a 70 kHz subcarrier, but they could be addressed digitally.