Just as it is important to be exact on what a pvalue shows and what it does not, it is equally important to be precise on what "replication crisis" means, and what it does not.
Specifically what it means is that the current binary system for describing significance is built precisely in such a way that even in the case of true effects, the probability that two studies would disagree could be as high as 50%. Therefore we need better metrics to describe the evidence. What it does NOT mean, is that all the research conducted is sham and worthless.
Specifically, if one study finds p=0.045 and another finds p=0.055, then the two studies will be in conflict (i.e. failure to replicate). But to use those pvalues to then claim "there is no evidence of an effect" shows a misunderstanding of the whole process, since a pvalue of 0.055 is actually pretty good evidence (just not as good as our arbitrary threshold).
Popularisation of the term "replication crisis" as a term is part of the problem if you ask me. It points the finger completely in the wrong direction.