I think the bandwidth part of the hack is the least interesting or important part of the hack. I am rarely, if ever, finding myself needing it if I am being honest. I always laugh hearing about the hobbyists chasing high bandwidth, 10 bit adcs, and fft for use with their Arduino projects. You just don't need it. 4 channels on the other hand, advanced trigger options and serial decoding, all very relevant for the hobbyist.
No amount of finesse can make up for inadequate bandwidth but it is often possible to get by with fewer channels especially if you make use of the external trigger.
Instead of thinking in terms of bandwidth, think in terms of transition time. 25 MHz TTL is about 3.5 nanoseconds which is 100 MHz for an oscilloscope. 100 MHz FAST and AS TTL is almost 1 nanosecond which is 350 MHz. Common low voltage CMOS logic is closer to the later than the former and no matter what the clock speed is, the fast edge is what matters and a 100 MHz oscilloscope can completely miss problems like double clocking even with the "slow" logic associated with an Arduino..
The point I was trying to make (and didn't actually state as such apparently) is that if you want the "best" scope (for YOU .. there is no best scope for everyone) you really need to figure out your requirements - I know this may not be easy for a hobbyist setting up a home lab, but you should be able to come up with a pretty good idea of the types of things you want to work on - from that determine your NEEDS and WANTS, do some research is pick the scope that best suits these parameters. If you just ask "what scope should I buy" in a forum like this, you will get information based on others peoples needs and biases.
It is easy to fall into the trap of "What if I need to see this someday" - one could argue that digital signals really want to be square waves, and in order to see every nuance of a square signal you really should get yourself an infinite bandwidth scope with infinite sample rate - then there would NEVER be an artifact that you would not be able to see, even as technology gets faster and faster.
Unfortunately such a scope is unobtanium and the ones that come closest to that goal are VERY far beyond the reach of most small companies and hobbyists and usually need some pretty serious budget approval even in a large company.
So clearly a compromise in in order.
Through the 70-90's myself and other colleagues worked on TTL systems not realizing the 10-20Mhz scopes we could afford were too slow to be useful. But we saw everything we needed to see and produced a boatload of reliable products.
Were these 100Mhz or even 25Mhz TTL? NOPE ... and I'd bet most of the signals hobbyists are driving out of their arduinos aren't either. As noted in my first post, if you are working on high-speed modern designs then yeah - you need uber-fast tools. But such design work is beyond the scope of most home-labs so planning (and spending) for it may not make sense.
I'm not saying YOU don't need a 100Mhz (or faster scope). Only YOU can determine what you actually NEED/WANT, and therefore only YOU can determine what is the "best" scope for YOU.
But be aware of the compromises. In chasing the holy-grail of bandwidth (and diverting your purchase dollars that way), you may end up spending a large about of your time "getting by" without capabilities and features that would make your work easier on a day to day basis - but only YOU can choose the right compromise.
Btw: If you're patient (or lucky) you might be able to come pretty close to the best of both world without spending a lot - My TDS380 cost me $30 ... a client was e-cycling it because "it had a bad input", so I brought it home and took a look - someone had replaced one BNC connector with a non-Tek part that swapped ground and probe sense .. a replacement cost me $30 and now it works perfectly.