Is it possible to combine a developer and a tester in one person?

Question

Personnel Management

Situation: A small company is developing its own project. 2 full-time developers, each is engaged in his own part of this project (1st - back office with cabinets, 2nd - website).
By complexity: Backoffice is about as complex as an average CRM system. Site - a regular site on Yii2 with a simple client cabinet (filling in your profile data). Requirements and wishes change frequently, automated testing and TTD are considered very expensive for us in terms of labor costs.
Primary testing is carried out by the developers themselves, then a third-party person is involved (NON-professional tester, but we will call him that). It is believed that the code from the programmer should be close to the release one, with minimal bugs, but in fact there are quite a lot of bugs.
Questions:
1) What is the normal number of bugs in the version that is submitted to the tester? How to define the boundary, what bugs should be found and fixed by the developer himself?
2) Can a developer independently test his project to the release state in order to get rid of the tester altogether? What are the arguments for? What are the arguments against?

Answer 1

> What is the normal number of bugs in the version that is submitted to the tester?
The number of bugs that the developer knows about should be 0. And those that the tester finds - this already depends on a bunch of factors. There is no such thing as a normal quantity.
> How to determine the boundary, what bugs should be found and fixed by the developer himself?
How to determine the limit, how many times do you need to change the specification?)))
> Can a developer independently test his project to the release state in order to get rid of the tester altogether?
No. First, because his gaze is blurred. Secondly, because the price of an hour of a programmer's work is higher than that of a tester.
This is the same as putting a surgeon at reception.

Answer 2

Requirements and Wishlist change frequently

automated testing is the very first thing that reduces the cost of development and the number of bugs.
who is considered? how much is the minimum, how much is enough?
some incomprehensible values, it is very difficult to understand what you generally mean and who installed such a quantity.
you need to write automated tests - they should all pass before release.
-------
1) Each feature must be covered by at least one test case, all tests must be passed. The number of bugs identified by the programmer himself should be equal to zero - he passes it on to another tester precisely because he cannot find them anymore.
2) Yes, maybe testers are not needed in small products.
Arguments for testers - sometimes it is difficult to find errors in your own code (biased view).
The argument against is extra coordination of work + stupid monkeys no matter how much they dug, they still won’t be able to competently test the product, to create good tests, good developers are needed who more or less understand about the completeness and equivalence of functions, can create test generators, etc.