Improve metadata in functions and component file

[veenstrajelmer]

TODO:

• two header examples in Pass metadata between objects #93
• first improvements in Get nodalfactors/xfac settings from component file #144
• how to make computer-readable (and writeable) analysis settings for component file (necessary since we will run with fu_alltimes soon)
• remove commented line in metadata_from_obj >> no, this would make using noos files or netcdf files way more complicated.
• remove metadata_from_ddlmeta(), not used/tested (also not in kwk repos) and seems not generic
• avoid writing diafile with tzone!=1 (already done). Mention somewhere "in DONAR zijn alle data in MET" (so prevent writing with other tzone should not be possible)
• avoid writing compfile with nodalfactors=False or fu_alltimes=True or source!="schureman" until the metadata can be included in header. Even then, we still cannot remove stats_xfac0 from example scripts since we need it when starting from diafiles
• make source (schureman/foreman) also attribute to be added and checked everywhere
• update metadata for extremes in calc_HWLW? At least TYP >> not necessary since we distinguish with HWLWcode columns in dataframe. Could be done with additional groepering attribute, but would make it too complex.
• prevent prediction with xfac=True with compfile with xfac=False (or vice versa). Compare all ana/pred options in prediction() function.

Non-parsed metadata from component file (example headers in #93):

• *valid for year
• *SA/SM from different period
• *theoretische f-factoren (means xfac=0)
• *component splitting
• CODE=3 >> not relevant for hatyan2

Missing metadata in component file:

• timestep_min (PERD start stop 60 is timezone according to anadea.f)
• nodalfactors on/off (always on)
• xfac (off if "theoretische" in headerlines)
• nodalfactors on all timesteps or per year (always per year up to now)
• MIDD replaced by slotgemiddelde

Missing metadata in diafile:

• timezone, but donar is always MET

Current status is:

* written with hatyan-2.7.2
* origin :  from timeseries dia file
* nodalfactors :  True
* xfac :  True
* fu_alltimes :  False
* groepering :  NVT
* timestep_min :  60.0
* timestep_unit :  min
* TYP :  TE
* components_sec :  SA, SM imported from component file
STAT  VLISSGN    WATHTE    NAP    cm    1
PERD  20090001  000000  20120031  230000     60
COMP      3
MIDD      1.00
NCOM    95
COMP    0     0.000000     1.000    0.00  A0          
[...]

MWE:

import os
import hatyan

dir_testdata = 'C:\\DATA\\hatyan_data_acceptancetests'
current_station = 'VLISSGN'

#read timeseries, components from analysis, write component file
file_data_comp0 = [os.path.join(dir_testdata, 'predictie2019', f'{current_station}_obs{i}.txt') for i in [1,2,3,4]]
ts_measurements_group0 = hatyan.readts_dia(filename=file_data_comp0, station=current_station)
comp_frommeas_avg_group0 = hatyan.analysis(ts=ts_measurements_group0, const_list='year')
hatyan.write_components(comp_frommeas_avg_group0, 'test_frommeas.cmp')

#read+write component file
file_data_comp1 = os.path.join(dir_testdata, 'predictie2019', f'{current_station}_ana.txt')
COMP_merged = hatyan.read_components(filename=file_data_comp1)
COMP_merged.loc['A0','A'] = 1.23 #TODO: add to metadata if overwritten
hatyan.write_components(COMP_merged, 'test_fromfile.cmp')