SCIAMACHY L2 to L3 with CODA?

CODA
2

This is possible. You would have to read the additional parameter with coda. And then add it to the HARP product before you perform any filtering.
And after combining the data you can perform the filtering.

For instance:

product = harp.import_product("SCI_OL__2PYDPA20060101_010721_000035612043_00475_20068_0000.N1", options="dataset=nad_ir1_ch4")
pf = coda.open("SCI_OL__2PYDPA20060101_010721_000035612043_00475_20068_0000.N1")
product.err0 = harp.Variable(coda.fetch(pf, "nad_ir1_ch4", -1, "non_linear_fit_param_err", 0), ["time"])
product.err1 = harp.Variable(coda.fetch(pf, "nad_ir1_ch4", -1, "non_linear_fit_param_err", 1), ["time"])
product.err2 = harp.Variable(coda.fetch(pf, "nad_ir1_ch4", -1, "non_linear_fit_param_err", 2), ["time"])
coda.close(pf)
filtered_product = harp.execute_operations(product, 'scan_direction_type=="forward";err0>=0')