PPPoE works by transferring PPP packages over Ethernet, as the name implies. This "over ethernet" is literal - your FTTH modem is a media converter and passive Ethernet switch and whatever device on your side which talks PPPoE talks with some device in a junction box of your provider. This would be a large gray box at the side of the street (for my house), or some box inside your building, belonging to your telco company, gathering all the connections from the individual flats.
Wifi is the same - there is no cable, but on a topology level, the Wifi is a media converter, and on top of it there sits regular Ethernet.
Also, in the usual home Wifi routers that I have experience with, the Wifi devices are by default (and often this cannot even be changed) on the same network as the physical LAN ports of the Wifi router.
Long story short, Ethernet is Ethernet is Ethernet. By connecting the FTTH modem to a LAN port of your Wifi router, and connecting all other devices in your home to the LAN port of your Wifi router, or adding Wifi devices, all these devices (or rather, network interfaces), including the PPPoE upstream of your provider, sit on the same Ethernet.
It does not matter that there are three different physical media here (the FTTH line; the Wifi; and the physical Ethernet cables you might or might not have). All these are Ethernet; all devices have MAC addresses, there is no routing on this level, everything can see everything (assuming no MAC filters in any of the devices, which there usually are, at least for even home-use Wifi routers, in my experience).
If you do nothing else, then all these devices can communicate with each other, all devices will specifically receive all broadcasts, and you can set up IP on top of it via your DHCP server, or static IP setup; but you have no Internet/upstream connectivity yet. This is where PPPoE comes in. Any device on your combined Ethernet can talk PPPoE with your provider. Your provider has no clue whether it is the FTTH modem, the Wifi router, or a little Raspberry Pi Linux box sitting in a corner.
Whatever is your PPPoE "client", will have a virtual or physical Ethernet interface as well. Traffic to/from the upstream Internet will go through this interface. In case you configure the PPPoE manually on a Linux box, you can see this interface in the list of interfaces; it is by default not the same interface as the LAN, and it is then up to you to set up routing (not switching) between your LAN interface and the PPPoE "Internet" interface (you would, for example, define the IP address of this interface as gateway in your DHCP or static configs; you would set up NAT there, and make sure that incoming TCP/IP connections are blocked, and so on and forth; all your security measurements go into this routing).
Finally, the WAN port of your Wifi router has the special semantics that if the Wifi router acts as PPPoE client, it will route the PPPoE connection through this port, and this port only; also this WAN port would usually not be switched with your LAN. This is a default configuration which works well in this scenario, as it separates your LAN Ethernet from the Ethernet that is used for the PPPoE traffic. By connecting the WAN and LAN ports on your Wifi router, you combine the two Ethernets into one. The most interesting implication is that some technician having access to the upstream device (the PPPoE server) could use a packet sniffer to see your LAN ethernet packages, or do further shenanigans (i.e., add his own devices to your LAN).
(Note that modern switches will, as soon as they know which of their LAN ports has a certain MAC address connected, will usually only send packages destined to that MAC on that port - so in practice the above scenario of the technician listening to your LAN would only happen if you have incredibly dumb, outdated switches; but always for Ethernet broadcast frames. Adding new LAN devices on their site should still be possible, you have to decide for yourself if this security implication bothers you.)
