TCRM version 3.1.3

More bug fixes:

• A bug where the latitude and longitude were read as doubles, not floats, which rounded the values that led to errors when sampling the mean sea level pressure data
• Location id values were strings, not integers which led to empty queries from the hazard database when plotting return period curves
• Add some verbosity for messages around selection of kernels in the statistical sampling components
• Fix the autocorrelation definition in tests
• Add a __call__ method to DummyStatus in the dummy mpi4py class, allowing the hazard module to complete in serial execution
• AutoPlotHazard.py now reads the number of simulations from the configuration file, rather than hard-coded 10000 simulations