Colour temperature and emission spectroscopy

At some point all we have are hypothetical models, but we can see which model seems to explain all the data we have so far. Already back in 1802 Fraunhofer observed absorption lines in the solar spectrum, and Kirchhoff and Bunsen later found that some of them correspond to emission lines in the spectra of heated elements. It was also known and shown by Foucault since 1849 that emission lines and absorption lines are the same for the same element, which we now know is because they correspond to energy levels of the particles that absorb or emit the photons (discovered by Einstein). At the same time, we have a model of black-body radiation that gives a reasonable fit for the sun at 5778K. Besides the spectral lines due to specific elements, there are many reasons why the fit is not super close, and all these are part of our current best model.

Moreover, it is difficult for an extremely large object to mimic a very different black-body spectrum. An element may have an emission line with a 'yellow' wavelength, but that is not at all the same as having an entire roughly continuous spectrum of a 'yellow' star. The colours you see from burning different elemental salts in a non-luminous bunsen flame arise from discrete spectral lines rather than black-body spectra.