My take, for what it is worth, is that one reason for embedding the input profile would be to retain all of the colour information that a sensor can capture.
...
The camera input profile might have its own limitations or foibles, depending on the camera in use and the quality of the icc profile, but as i understand it, it gives you the best shot at retaining all that the sensor has to offer.
not quite right.
C1 works with a certain internal color space (apparently a huge color space). The input profiles ("camera" profiles) are assigned to the files in order to create a certain look. Some camera profiles are quite large, some are quite small (depending on the camera model). The profiles are not referring to the colors the respective camera actually can capture. Actually it would be appropriate to call the input profiles "camera working spaces".
But... those input profiles are assigned anyway at the beginning. Therefore there is no gain in a further conversion to a particular color space (AdobeRGB, ProPhotoRGB or whatever). For instance: when the input profile is much smaller than ProPhotoRGB but you set your output to ProPhotoRGB... then you are only transferring the very same colors in a larger container (which might be useful with regard to workflow reasons... but not with regard to preserving/discarding colors).
I do embed the "camera" profiles and also leave the files in their "camera" profiles when adjusting the files in Photoshop and only convert the final file into a certain output profile (printer profiles or sRGB for web).
As to the issue: yes, the Color Editor alters the actual input profile. When you add a Color Editor layer there is no way to merge the two profiles (the original input profile on the bottom layer and the altered profile on the upper layer) into one new profile. Therefore C1 chooses the smallest nominator as output profile (which is sRGB).
When you work with Color Editor layers you simply should set a reasonable output profile of your choice (AdobeRGB, ProPhotoRGB, ECI-RGB... whatever).