An MVP is not a product, it's an MVP

tl;dr: Build, measure and learn. But don't avoid best practices afterwards.

Minimum Viable Product, or even better Minimum Viable Prototype (MVP from now on), is a technique born in the Lean Startup movement to reduce the risk of gaining validated learning, without spending too much effort, money and time in the wrong project direction.

You build a MVP to validate an idea, be it for a brand new product or a new feature in an existing one. In this stage of the project you don't care about: tests, QA, copy-pasting code, mocking the UI, security, code reviews, maintainability, scalability, etc.

In case this prototype is successful, you should go and build the product or the feature to capitalize that validated learning, but this time, following the industry's best practices, so it can be used in production by your (paying?) users.

This last step is crucial. There will be the temptation of not taking it. The consequences of not doing it are well known, that's why we have best practices afterall. The risk is low now. Money, time and effort won't be wasted. There is no excuse. An MVP is not a product, it's an MVP.