This is the common problem with using CERA like this. Catchers catch different pitchers at higher rates than others, so you have to look at each pitcher individually. Of course, some sample sizes are very small, especially, if one catcher hardly ever catches a certain pitcher, which is often the case.
I have done exhaustive studies on this, pitcher by pitcher. Leon blew Vaz away in about 75-80% of the pitchers with large enough sample sizes, both ways, to go by. Before that, all other back up catchers, except Swihart also had better numbers than Vaz. When it happens over and over, it's likely not a fluke.
I have not done Plawecki v Vaz in a while, but here it is...
Eovaldi:
2021: 2.66 Wong (20 IP), 4.30 Vaz (58 IP),4.72 Plawecki (48 IP)
2020: 0.64 Plawecki (14), Vaz 4.98 (34)
Career: 2.66 Wong (20), 3.79 Plawecki (62), 4.30 Vaz (147), 4.53 Leon (44)
ERod:
2021: 5.33 Vaz, nobody else has caught him, this year.
2020: n/a
Career: 3.78 Hanigan (50), 4.05 Leon (118), 4.24 Vaz (560), 4.44 Swi (75)
Pivetta: Career: 3.09 Plawecki (32), 4.48 Vaz (96)
Richards: 2021: SSS 3.60 Plawecki (15), 5.48 Vaz (95)
Perez: Career 4.14 Vaz (146), SSS Plawecki 9.35 (17)
Barnes: Career 3.44 Vaz (196), 3.83 Leon (99), SSS 6.94 Plawecki (12)
Looks like mixed results and many lop-sided sample sizes, but still tilts towards the back-ups over Vaz.
I don't have time to do every RP'er, and their sample sizes are even smaller.