[v7r1] using MixedEncode instead of Encode

[fstagni]

It looks OK IMHO, but I didn't test it.

BEGINRELEASENOTES

*Core
CHANGE: Using MixedEncode instead of Encode everywhere.

ENDRELEASENOTES

closes #4376

[chaen]

that will probably not work just like this. We originally said that we want to replace only the network serialization, not the dumps in the dbs and the rest, so that part was not tested, nor did we asses whether the db fields would be long enough etc. So if we really want to go that way, I'd put that in v7r2, not v7r1

[fstagni]

After discussion in https://indico.cern.ch/event/1008133/ we close this PR, because it's deemed "too risky".
Conclusion is that we can live with DEncode for decoding/encoding things that go to the DB.