Improved database upgrade experience

Currently if you hit an issue updating the schema, you'll get a non-specific error like this:

```
[ERROR] One or more fatal errors occurred
[ERROR] Unable to connect to the dependency-check database
```

If you rerun it with e.g. `dependency-check -log /dev/stdout -s .` part of the output will include more details:

```
org.owasp.dependencycheck.data.nvdcve.DatabaseException: unable to upgrade the database schema - please run the dependency-check purge command to remove the existing database
at org.owasp.dependencycheck.data.nvdcve.DatabaseManager.updateSchema(DatabaseManager.java:422)
at org.owasp.dependencycheck.data.nvdcve.DatabaseManager.ensureSchemaVersion(DatabaseManager.java:491)
at org.owasp.dependencycheck.data.nvdcve.DatabaseManager.initialize(DatabaseManager.java:214)
at org.owasp.dependencycheck.data.nvdcve.DatabaseManager.(DatabaseManager.java:123)
at org.owasp.dependencycheck.data.nvdcve.CveDB.(CveDB.java:238)
at org.owasp.dependencycheck.Engine.openDatabase(Engine.java:987)
at org.owasp.dependencycheck.Engine.doUpdates(Engine.java:863)
at org.owasp.dependencycheck.Engine.initializeAndUpdateDatabase(Engine.java:680)
at org.owasp.dependencycheck.Engine.analyzeDependencies(Engine.java:606)
at org.owasp.dependencycheck.App.runScan(App.java:254)
at org.owasp.dependencycheck.App.run(App.java:186)
at org.owasp.dependencycheck.App.main(App.java:81)
2021-06-10 14:13:34,650 org.owasp.dependencycheck.utils.WriteLock:240
DEBUG - Lock released (main) 35a3b24da5b9b9b1a6369aa1eddc029a @ 2021-06-10 14:13:34.65
[ERROR] One or more fatal errors occurred
```

That would give the impression that you could run "dependency-check purge" to purge the database but that will give an error. The `--help` output has nothing about purging but if you look at the README there is mention of the otherwise undocumented "--purge" option.

There are a couple of improvements which could make this smoother:

1. Include the error message in the standard output level — `DatabaseException` should always be exceptional so it seems like there'd be no downside to always emitting an ERROR-level message for those.
2. Add `--purge` to the CLI help
3. Simply attempt to purge the database automatically when this happens — it doesn't seem like any database instance should ever have irreplaceable information and anyone who encounters this is going to need to delete the database at some point.