nxbincache only purged when server is restarted

1. install Nuxeo
2. install S3 addon
3. create a File document and upload a file to it
4. check the tmp folder
5. observe an nxbincache folder is present
6. wait for some time, even an hour
7. observe the file is still present
8. stop Nuxeo
9. observe the file has been purged from tmp folder

Expected behavior: in production, some environments cannot be restarted on the sole purpose of cleaning the tmp folder. This should be done on-the-fly.
Note: this is due to the nxbincache file being created as a JVM temporary file, thus deleted only when the JVM stops:
https://github.com/nuxeo/nuxeo/blob/master/nuxeo-core/nuxeo-core-api/src/main/java/org/nuxeo/ecm/core/blob/binary/CachingBinaryManager.java#L105
using:
https://github.com/nuxeo/nuxeo/blob/master/nuxeo-runtime/nuxeo-runtime/src/main/java/org/nuxeo/runtime/api/Framework.java#L612

Also, the folder is declared as a LRU cache but checking the getSize method defined here:
https://github.com/nuxeo/nuxeo/blob/master/nuxeo-common/src/main/java/org/nuxeo/common/file/LRUFileCache.java#L136
it is never used except in testLRUFileCache:
https://github.com/nuxeo/nuxeo/blob/master/nuxeo-common/src/test/java/org/nuxeo/common/file/TestLRUFileCache.java#L58

So the max size is never evaluated.