That is also what I originally thought too. But I don't observe the number of followers of rival religions being halved when I spread the religion, even in a City State which would not have any capability to "resist" conversion/erosion as far as I am aware.
It doesn't work as you think. There are two concepts: pressure pool and pressure. The pool is the accumulated pressure points in the city, while pressure is just how many potential pressure points are exerted to the city and beyond.
Suppose a city with a pantheon and 0 religion points, only pantheon points. You spread once with a missionary, which adds 1000 points of his religion to the city. Pantheonists can add points to the city, but they can't resist conversion, so suppose the pantheonists have added 500 points up to this moment, in a 9 pop city. Once you add 1000 (let's say hinduist) points, the city has 1500 religious points (1000 hinduist + 500 pantheonist), so proportionally it is 66% hinduist and 33% pantheonist, which makes 6 hinduist citizens and 3 pantheonist citizens in the city.
After the spread, this city is going to exert 6 hinduist pressure to nearby cities (but it decays with steps traveled).
You spread a second time with the same missionary, giving another 1000 hinduist points, making the proportion 2000 to 500, so 80% hinduist, 20% pantheonist, which makes 7 hinduist citizens and 2 pantheonist citizens in your 9 pop city.
Now the city exert 7 hinduist pressure.
Let's say that some turns have passed. Your city is now 16 pop, all of them are hinduist, and your city has accumulated 10000 hinduist points. Here comes a portuguese catholic missionary and spread to your city, twice, adding 1750 catholic points to the pool. That's 85% hinduist and 15% catholic, meaning 13 hinduist and 3 catholic in the city. Not a great achievement, eh?
Then comes the inquisitor. Portugal conquers the city and send an inquisitor. The inquisitor removes all hinduist points in the city, leaving only 1750 catholic points in it, so the action causes 100% citizens in the city to be catholic, at least during this turn. Next turn the city is going to add all the religious pressure from the neighboring cities and things may change.
Suppose that we had built a mandir in the city, giving 25% conversion resistance just before Portugal conquered the city. When Portugal tried to remove heresy, instead of removing 10000 hinduist points, it would have removed only 7500 points, leaving still 2500 hinduist points in the city. Thats 62.5% hinduist, 37.5% catholic, so 10 hinduist citizens and 6 catholic ones, not being sufficient to convert it.
I guess that the ability for the missionaries eroding existing (accumulated) pressure is as follows:
In the case an inquisitor had removed 10000 hinduist points, the missionary removes 5000 points. In the case where the inquisitor had removed 7500 points, the missionary removes 3750. And then adds 1000 catholic pressure.
I can be wrong with the actual implementation, though.