I've never done any flatbed scanning personally, as I got on board with DSLR scanning as soon as I started using film (about 4 years ago now). From numerous third party accounts and comparisons I've read though, with good technique DSLR scanning easily surpasses the quality achievable with consumer flatbeds, especially for 120 and 35mm. It takes a bit more tinkering and work, perhaps, but I think the results are worth the effort. The stitching process is generally no problem, and is handled automatically by Lightroom (or Photoshop). I've found it is best to not refocus the lens between capturing each segment, as even small variances in focus create very obvious transitions when the frames are stitched together. For best results each segment should also have lens distortion corrections applied before stitching.
The only hassle I've encountered is sometimes the stitching can struggle if it has to join segments together containing almost no detail (large patches of blank sky, for example), as the algorithm then can't reliably determine reference points between frames. Usually just capturing more frames can get around this problem though.
Eventually I would like to get a mirrorless camera capable of high resolution pixel-shift captures (the Sony A7R IV, for example), which would obviate the need for multi-frame stitching. This would make the whole process much faster whilst retaining the same (or higher) quality, assuming one used a good enough lens.
Bookmarks