With some regular testing with each new Vault version to update the parts of the Consul backend that are and are _not_ replicated, this may be worth trying.
Did it work well for you?
Did it turn into a "penny-wise, pound foolish" situation?
Something in between?
Inquiring minds want to know.
There are a few results on consul-replicate from 2016, and the following are negative, but only apply to some points:
https://groups.google.com/forum/#!searchin/vault-tool/consul-replicate|sort:date/vault-tool/3Ti37k1G4V4/sebvuDiLCQAJ (Multi-master replication with consul-replicate would create conflicts, but doesn't indicate that Secondaries would fail)
https://groups.google.com/forum/#!searchin/vault-tool/consul-replicate|sort:date/vault-tool/fji21P13SsU/72ex89X2AAAJ (Replication would require replicating the encryption keys)
And here's a post that indicates that consul-replicate as a means of Vault performance replication (e.g. replication of data across multiple datacenters for more local availability) is unsafe "in nearly all circumstances":
https://groups.google.com/forum/#!searchin/vault-tool/consul-replicate|sort:date/vault-tool/RMpbRGSq2A8/zRZP_xFfAwAJ (consul-replicate for Vault replication as a way to do multi-DC availability is unsafe in general)
Anyway, I did a bit more of my own research and testing. Doesn't seem like it's a popular solution due to potential data risks / scaling issues