boards/nucleo-f429zi: add support for ethernet

[aabadie]

Contribution description

As the title says. With #17827 which is very similar, I saw that the Ethernet configuration was missing for this board.

Testing procedure

The Ethernet interface pops up in ifconfig when flashing examples/default:

2022-03-18 16:10:12,993 # NETOPT_RX_END_IRQ not implemented by driver
2022-03-18 16:10:12,996 # main(): This is RIOT! (Version: 2022.04-devel-893-g1de68-pr/boards/nucleo-f429zi-eth)
2022-03-18 16:10:12,996 # Welcome to RIOT!
> 2022-03-18 16:10:13,911 # PKTDUMP: data received:
2022-03-18 16:10:13,912 # ~~ SNIP  0 - size:  96 byte, type: NETTYPE_UNDEF (0)
2022-03-18 16:10:13,923 # 00000000  60  00  00  00  00  38  00  01  FE  80  00  00  00  00  00  00
2022-03-18 16:10:13,929 # 00000010  DE  82  33  B9  9A  73  C6  90  FF  02  00  00  00  00  00  00
2022-03-18 16:10:13,935 # 00000020  00  00  00  00  00  00  00  16  3A  00  05  02  00  00  01  00
2022-03-18 16:10:13,947 # 00000030  8F  00  32  B2  00  00  00  02  04  00  00  00  FF  02  00  00
2022-03-18 16:10:13,948 # 00000040  00  00  00  00  00  00  00  00  00  00  00  FB  04  00  00  00
2022-03-18 16:10:13,953 # 00000050  FF  02  00  00  00  00  00  00  00  00  00  01  FF  73  C6  90
2022-03-18 16:10:13,959 # ~~ SNIP  1 - size:  20 byte, type: NETTYPE_NETIF (-1)
2022-03-18 16:10:13,960 # if_pid: 3  rssi: -32768  lqi: 0
2022-03-18 16:10:13,960 # flags: 0x0
2022-03-18 16:10:13,965 # src_l2addr: D8:EB:97:BF:2F:A7
2022-03-18 16:10:13,965 # dst_l2addr: 33:33:00:00:00:16
2022-03-18 16:10:13,971 # ~~ PKT    -  2 snips, total size: 116 byte
2022-03-18 16:10:14,146 # PKTDUMP: data received:
2022-03-18 16:10:14,151 # ~~ SNIP  0 - size:  96 byte, type: NETTYPE_UNDEF (0)
2022-03-18 16:10:14,157 # 00000000  60  00  00  00  00  38  00  01  FE  80  00  00  00  00  00  00
2022-03-18 16:10:14,165 # 00000010  DE  82  33  B9  9A  73  C6  90  FF  02  00  00  00  00  00  00
2022-03-18 16:10:14,170 # 00000020  00  00  00  00  00  00  00  16  3A  00  05  02  00  00  01  00
2022-03-18 16:10:14,176 # 00000030  8F  00  32  B2  00  00  00  02  04  00  00  00  FF  02  00  00
2022-03-18 16:10:14,188 # 00000040  00  00  00  00  00  00  00  00  00  00  00  FB  04  00  00  00
2022-03-18 16:10:14,190 # 00000050  FF  02  00  00  00  00  00  00  00  00  00  01  FF  73  C6  90
2022-03-18 16:10:14,194 # ~~ SNIP  1 - size:  20 byte, type: NETTYPE_NETIF (-1)
2022-03-18 16:10:14,199 # if_pid: 3  rssi: -32768  lqi: 0
2022-03-18 16:10:14,200 # flags: 0x0
2022-03-18 16:10:14,200 # src_l2addr: D8:EB:97:BF:2F:A7
2022-03-18 16:10:14,205 # dst_l2addr: 33:33:00:00:00:16
2022-03-18 16:10:14,212 # ~~ PKT    -  2 snips, total size: 116 byte
ifconfig
2022-03-18 16:10:18,111 # ifconfig
2022-03-18 16:10:18,117 # Iface  3  HWaddr: A6:D7:5E:88:84:3E  Link: up 
2022-03-18 16:10:18,123 #           L2-PDU:1500  Source address length: 6
2022-03-18 16:10:18,123 #           
> 2022-03-18 16:10:19,637 # Exiting Pyterm

Issues/PRs references

Noticed that when working on #17827

[maribu]

ACK. Code looks good and testing is provided

[maribu]

TODO for myself: Check if this is actually still used and drop for all boards consistently if not. If I recall correctly, the previous version of the driver used this to speed up memcpy to a TX buffer. But I rewrote the driver to use the ETH DMA to directly collect the data to transmit in a zero-copy fashion. If I recall correctly, the DMA stream can be used for other stuff, as the ETH DMA is a distinct DMA unit.

[aabadie]

Thanks !