Because it's a sample issue. I'm not going to go on a deep dive regarding an issue that is common sense.
CERA:
1) Doesn't control for pitcher quality.
2) Doesn't control for standard deviation because of the varying sample sizes.
3) Doesn't account for quality of defense, which may vary on any given night.
4) Doesn't account for umpire tendency, which can greatly impact CEra on smaller samples.
5) Doesn't account for park factors.
There are more advanced stats that measure the impact of all of this factors, one way or the other, for pitchers. Even more advanced metrics like catcher framing are inherently flawed. You can't use those as gospel, because they have fundamental flaws, so why would you use CEra, a worse, less complete statistic, to measure a catcher's worth? I jus don't see it.