The new implementation allows not to provide database names
in the connection strings. The connection string parser was
extended with an additional parameter allowing to privide
the default database name if the one is found missing in
the input connection string.

The change was made to the Configuraton servive of the Replication/Ingest
system.

Refactored dependencies accordingly. The move allows re-using the
class by other Qserv modules w/o adding a direct depedency onto
the module 'replica' which was the original location of the class.

Fixed a bug in the implementation of the timer to allow subsequent
restarts. Unblocking implementation of the callback calling mechanism.
Extended a model of the callbacks to allow ordering automatic restarts
of the timer via the return value (boolean type) of the callbacks.
Minor refactpring of the implementation.

Extend the Histogram class to report the histogram label (id).
Added a histogram to capture the performance of the result file
reads at Czar. Added counters for the number of on-going
result file reads (file-based result delivery protocol only)
and merges. Fixed database merge histogram load.
Added integral counters for metrics reset at the start up time
of Czar. Added the timestamp (milliseconds) attribute to the JSON
object that returns a state of the monitoring parameters. The timestamps
are going to be used for the performance monitoring of Qserv.

The history is recorded in the transient monitoring store
(class qserv::CzarStats) from where it could be sampled
at run time via the Qserv mysql-proxy service.

The requests are made by calling a special stored procedure
that is recognized by Qserv. The procedure allows a single parameter
(a command) that is interpreted by Czar and processed accordingly.

The fix was introduced after discovering that the shared connection
usage approach in the original design of this and related classes
is highly unreliable. A connection shared by the queries seems
to be sensitive to the service startup ordering and occasional
disconnects. As a result of that the shared connection could just
stop working or it could not be alive from the very start of Czar.
It's also possible that the shared connection object is not properly
synchronized to allow using it in the multi-threaded environment
(which would be a bug). The last theory will be further investigated later.

The service was replaced with two separate services. The first one
reports info on the ongoing queries. The second service is used for
searching and displaying info on the past queries.

The improvement aimed at reducing code duplication as well as the total
amounbt of code in the application.

Separated the query into into two pages - the active queries and
the past queries. Also rearranges sub-tabs under the 'Status' tab
for more intuitive navigation.

Added a link (an additional column on the active queries table) to
the newely added Query progression plot.
Fixed minor layout bugs on the active and past query tables.
Loading the Highcharts.js library on demand.