[Performance] Excel-like filter for huge data sets

[fipro78]

There are several places in the code that is used for the Excel-like filter dropdowns, that contain possible performance flaws for huge datasets.

• NatCombo
  • setDropdownSelection(String[]) - inside a for-loop a contains() operation is executed to calculated the selectionStateMap and the checkbox state. This should be replaced with a Map based operation.
  • getSelectionIndices() - inside a for-loop an indexOf() operation is executed to calculated the selected indexes based on the items. This should be hcanged to a Map based operation.
  • The Integer usage should be replaced with primitive collection usage.
• FilterNatCombo.getTransformedText(String[])
  Use a StringJoiner instead of + operator for String concatenation
• ComboBoxGlazedListsFilterStrategy.getStringFromColumnObject(int, Object)
  Use a StringJoiner instead of + operator for String concatenation
• FilterRowComboBoxCellEditor.setCanonicalValue(Object)
  The calculation of diffValues is using removeAll() to calculate the diff between currently visible and all available items. To improve the performance we should change the type of diffValues from List to HashSet as the order doesn't seem to be relevant.