XAS: Fix Conflict Between tot_charge Override and CH Protocol (#809)

- Fixes a conflict between `tot_charge` being set explicitly in overrides by default and the XAS plugin assuming that `tot_charge` isn't specified in the overrides.
- Adds a note to the documentation for the XAS plugin regarding charged systems and provides some suggestions for how to handle them.

docs/source/howto/xas.rst

@@ -167,6 +167,19 @@ Finally, click on the "Download CSV" button to the upper left of the plot area t
     * The normalised & weighted spectrum. (with respect to ratio of site multiplicity to total multiplicity)
     * The normalised & un-weighted spectrum.
 
+Additional Note on Charged Systems
+----------------------------------
+Computing XANES spectra for systems where a charge has been applied (in this case using the `Total charge` advanced setting) is possible using the tools
+available in the QE App, however such workflows should always be tested by the user against experimental data if possible.
+
+When running XAS workflows for systems where a total charge has been applied, it is suggested to use the following settings for the total charge to ensure the corresponding
+core-hole treatment is applied correctly:
+
+* "xch_fixed" or "xch_smear": Set the total charge as required for the system's charged ground-state.
+* "full": **Increase** the total charge by 1 *relative* to the system's charged ground-state (e.g. set total charge = 2 in the advanced settings tab if the charge is normally 1).
+
+Note that for neutral systems (total charge = 0), the QE App will handle these settings automatically.
+
 Summary
 -------
 

src/aiidalab_qe/plugins/xas/workchain.py

@@ -23,6 +23,11 @@ def update_resources(builder, codes):
 def get_builder(codes, structure, parameters, **kwargs):
     from copy import deepcopy
 
+    adv_parameters = deepcopy(parameters["advanced"])
+    # Setting `tot_charge = 0` will cause FCH calculations to fail due to
+    # inputs being incorrect, thus we pop this from the overrides
+    if adv_parameters["pw"]["parameters"]["SYSTEM"].get("tot_charge") == 0:
+        adv_parameters["pw"]["parameters"]["SYSTEM"].pop("tot_charge")
     protocol = parameters["workchain"]["protocol"]
     xas_parameters = parameters["xas"]
     core_hole_treatments = xas_parameters["core_hole_treatments"]
@@ -61,7 +66,7 @@ def get_builder(codes, structure, parameters, **kwargs):
     xs_code = codes["xspectra"]["code"]
     overrides = {
         "core": {
-            "scf": deepcopy(parameters["advanced"]),
+            "scf": adv_parameters,
             # PG: Here, we set a "variable" broadening scheme, which actually defines a constant broadening
             # The reason for this is that in "gamma_mode = constant", the Lorenzian broadening parameter
             # is defined by "xgamma" (in "PLOT"), but this parameter *also* controls the broadening value