It looks like it is not cost related, it looks like deliberate battery degradation ... and now that you mentioned cost here is example from a cheap Xiaomi (redmi note 2).
Once charged it does not use the battery.
You've discovered the difference between what the industry calls a "battery fed" design vs a "charger fed" design. Here's a good explanation of the basics:
https://www.analog.com/media/en/reference-design-documentation/design-notes/dn336f.pdfWith the "series" or "battery fed" design (which the Samsungs seem to have), the load supply is taken from the battery, so in your use-case the charger detects the battery is not full and begins charging, then shuts off once the battery is full --- allowing the load to discharge the battery below the "begin charging" point, and the cycle repeats. This is a cheaper topology since there's no real switching of power taking place; just one charger.
With the "parallel" or "charger fed" design, the load is connected directly to the power input and the battery through a charger/discharger. The power input both charges the battery and powers the load. Once the battery is charged and VBUS > VBAT the charger turns off and the load is directly powered from the input. The Xiaomi you have uses the common MTK platform, and (I also have an MTK phone, not a Xiaomi however, so can confirm) the PMIC on those appears to be based on the charger-fed topology.