LZ4HC should benefit us with a decompression speed improvement by a factor of 5 than zlib but with a slight worse compression factor. And decompression seems a bottleneck in reading TTree at least for files at a local disk.
We have an extensible framework to support additional compression algorithm. If you propose a patch hooking this compression algorithm, we could then test it on real use case and decide whether to include it or not in the distribution (in general we prefer not to increase dependencies and complexity unless there is a significant benefit).