I don’t think there is one approach that works for everybody. Regarding this comment: “This could create unnecessary resource problems e.g. on UAD cards” - WaveLab 9 now offers dynamic plugin processing which means that UAD DSP resources are only used when a clip with UAD inserted is either playing or being rendered. You can find the setting in the preferences>plugins>organize area.
You can choose each plugin you’d like to be processed dynamically.
I normally don’t master 100% in WaveLab but when I do, I find that there are plenty of insert slots. If a single clip needed more than 10 plugins, I would do some preliminary destructive processing to correct the big problems of why it would need 10 plugins in the first place. You can always do a “Save As…” of the audio file/clip to get back to your start if needed for any destructive processing. “Replace Audio File” in the montage makes it easy to do this.
For most normal album mastering, I do the initial processing in Pro Tools because I prefer that DAW workflow better. I find it easier to manage having slightly different plugin chains per song and routing through the analog gear (with an unprocessed mirrored copy to another input on my monitor controller) as well as spot cleaning up sections after analog capture with RX5 De-Click/De-Noise and other plugins much easier than in WaveLab. The use of Playlists and AudioSuite in Pro Tools make it easier in my opinion than trying to do all this in WaveLab.
Then after all the songs are initially processed through plugins and my analog chain, and then cleaned up, trimmed up and properly named, I load them into WaveLab to sequence the album (maybe adding additional fade outs and crossfades), add CD-Text/metadata, and do any final processing as needed on a clip (not regularly but sometimes I do a small tweak after analog capture to a song or two), and then apply the final limiting/dither in the montage output effects section. I never use the global master section in album masters because I don’t like to remember to save and recall it. Too much room for user error.
All of my captured and trimmed/cleaned audio from Pro Tools is at 32-bit float/96k or 88.2k and I downsample to 44.1k with Weiss Saracon, Myriad or RX5 for initial client approval via DDP. But, the new “Custom Montage Copy” makes it very easy to recreate your montage at a higher/different sample rate as needed after client approval.
The best part about this workflow is that the montage is not doing a lot of processing so it’s quick to render a DDP or WAV files of each track.