In my reading, "rarity" doesn't really reflect rarity at all, it reflects supposed power level and the character level where it can be expected to appear. Common items at level 1-4, uncommon at 5-9, rare at 10-14, very rate at 15-20. The pricing rules in the DMG are based on the idea that you get ten times the loot at each tier and thus the cost is also multiplied by ten. But the rarity ratings are deeply flawed, with flight items being a great example - more rare in no way makes for better flight. Rarities are also setting specific. That the flying broom is just uncommon makes perfect sense in a Hogwartsesque setting, but is in no way balanced against other flight items.
If your reading results in the text being deeply flawed, have you considered you are reading it wrong?
If common/uncommon/rare/very rare/legendary actually also reflected how common the item is to be found, and not just its power level, wouldn't that erase the deep flaw?
Like, winged boots being overly common is ok, because
they only arrive if the DM picks them. If the DM is using random tables, then the odds enough winged boots arrive for an entire party at low levels is low. So even then, getting a broom of flying at the next encounter won't be a disappointment, because someone will still want it.
And the carpet of flying? Likely to be desired when it arrives, because the odds the entire party has flight is low.
Heck, if you use random treasure tables, the odds the same party encounters two of winged boots, a carpet of flying, and a broom of flying and wings of flying is very low. So the comparison doesn't actually matter in a sense!
If the item's arrival in a random treasure parcel after the PCs have gotten previous random treasure parcels (with the occasional custom item) is exciting, the item is still a good reward for its rarity level. It being better than an item on another treasure table doesn't matter in actual play.
This problem only occurs if you open the treasure tables up as a shopping list.
And above, I talked about that. If you are going to have more magic items for sale, as a DM you should pick them. They can be a broad set of items. And you should consider what the impact is of items you add.
Only if you open up the
whole list, or pairs of items without considering their relationship, does this pricing problem occur.
Don't open up a magic mart, be it a catalogue service (physical or not) or a single store with a huge inventory, and the entire pricing problem evaporates.
If you are playing in Eberonn?
What I'd do is do some treasure table rolls for your version of Eberonn. Roll a bunch of times. Assign each roll to a house. Apply a variable markup to each item over the base DMG numbers (2x to 10x) for public purchase, with membership having its privileges (killing the markup).
Vary each of the rolls. So if you roll a +2 scythe, maybe say "ok, maybe not a scythe" -- except, maybe it is? Maybe an guide sells a kick as farming implement that acts as a +2 scythe. And that a good way to get yourself a high quality magic blade is to buy the farming implement and see if you can jury-rig the enchantment into a different weapon type; this is akin to how, in our world, the best way to make some kinds of tools is to buy something mass-produced for a different purpose and modify it.
Include common items for sale. Maybe roll some common items randomly and invent some variations.
There; adventuring magic items for sale, but not an infinite catalogue. Most magic items for sale won't be adventuring gear after all; they'll be items to make the economy run.
Once you have the general inventory, individual magic shops in a particular town can use the magic item tables. As manufacture of items in Eberonn is still somewhat "home workshop"y and craft based, a one-off item for sale is possible. Like, you can buy the journeyman items produced by the trainees (no warranty express or implied) which in turn helps pay for that student's education.
