Use the Retryable API in MongoDB connector

Currently Nuxeo is not robust when this scenario happens:

1. Configure Nuxeo with MongoDB HA with replicaset mode (1 primary node, 3 secondary nodes, 1 arbitrary node), as recommended by Nuxeo
2. Start Nuxeo
3. Run a program which creates 50000 documents - it should last at least several minutes
4. During the program execution, shut down the primary node
5. Observe that there are some connection errors

   Caused by: javax.net.ssl.SSLException: Connection reset
    at sun.security.ssl.Alert.createSSLException(Alert.java:127) ~[?:1.8.0_272]
    at sun.security.ssl.TransportContext.fatal(TransportContext.java:324) ~[?:1.8.0_272]
    at sun.security.ssl.TransportContext.fatal(TransportContext.java:267) ~[?:1.8.0_272]
    at sun.security.ssl.TransportContext.fatal(TransportContext.java:262) ~[?:1.8.0_272]
    at sun.security.ssl.SSLSocketImpl.handleException(SSLSocketImpl.java:1303) ~[?:1.8.0_272]
    at sun.security.ssl.SSLSocketImpl.access$300(SSLSocketImpl.java:72) ~[?:1.8.0_272]
    at sun.security.ssl.SSLSocketImpl$AppInputStream.read(SSLSocketImpl.java:831) ~[?:1.8.0_272]
    at com.mongodb.internal.connection.SocketStream.read(SocketStream.java:89) ~[mongo-java-driver-3.8.1.jar:?]
    at com.mongodb.internal.connection.InternalStreamConnection.receiveResponseBuffers(InternalStreamConnection.java:554) ~[mongo-java-driver-3.8.1.jar:?]
    at com.mongodb.internal.connection.InternalStreamConnection.receiveMessage(InternalStreamConnection.java:425) ~[mongo-java-driver-3.8.1.jar:?]
    ... 155 more
    Suppressed: java.net.SocketException: Broken pipe (Write failed)
    at java.net.SocketOutputStream.socketWrite0(Native Method) ~[?:1.8.0_272]
    at java.net.SocketOutputStream.socketWrite(SocketOutputStream.java:111) ~[?:1.8.0_272]
    at java.net.SocketOutputStream.write(SocketOutputStream.java:155) ~[?:1.8.0_272]
    at sun.security.ssl.SSLSocketOutputRecord.encodeAlert(SSLSocketOutputRecord.java:81) ~[?:1.8.0_272]
    at sun.security.ssl.TransportContext.fatal(TransportContext.java:355) ~[?:1.8.0_272]
    at sun.security.ssl.TransportContext.fatal(TransportContext.java:267) ~[?:1.8.0_272]
    at sun.security.ssl.TransportContext.fatal(TransportContext.java:262) ~[?:1.8.0_272]
    at sun.security.ssl.SSLSocketImpl.handleException(SSLSocketImpl.java:1303) ~[?:1.8.0_272]
    at sun.security.ssl.SSLSocketImpl.access$300(SSLSocketImpl.java:72) ~[?:1.8.0_272]
    at sun.security.ssl.SSLSocketImpl$AppInputStream.read(SSLSocketImpl.java:831) ~[?:1.8.0_272]
    at com.mongodb.internal.connection.SocketStream.read(SocketStream.java:89) ~[mongo-java-driver-3.8.1.jar:?]
    at com.mongodb.internal.connection.InternalStreamConnection.receiveResponseBuffers(InternalStreamConnection.java:554) ~[mongo-java-driver-3.8.1.jar:?] 

There are several leads to improve Nuxeo

• use the Retryable API (retryable reads and writes https://docs.mongodb.com/manual/core/retryable-writes/ and https://docs.mongodb.com/manual/core/retryable-reads/ )
• implement a mechanism similar to the "validationQuery" in JDBC resource

Expected behavior:

All documents are created without any loss. The possible errors are logged as warning if Nuxeo was able to recover the query and run it again without another exception.