Can´t really say that I have seen mentioned that it would have been voluntarily. Winkler had bit problems at the ending for previous season and the stories told that they brought Tiny Thompson as understudy to Winkler if he starts to slip again. And in "training camp" Thompson convinced Ross. Though I found one of my bookmarks where it is mentioned that Winkler "unselfishly helped" Thompson to get in big league so maybe?
The Boston Globe November 26,1928
The Calgary Daily Herald - Nov 29, 1928
Now but in here it need to be mentioned that Thompson wasn´t some rookie nobody. He had been targeted by pro clubs years before, but decided to stay as "amateur" in USAHA/AHA. He was star in league which was a pretty strong at the time. I guess another example how difficult it was to lure young talent to openly pro leagues at the time.
The Winnipeg Evening Tribune, 1924-08-29
Fairly certain that Winkler did not give up his crease voluntarily. Initial reports that he had "quit" the organization altogether after being cut were quashed by Art Ross, who made it clear that Winkler would be loaned to AHA Minneapolis which was closely aligned with Boston at the time. Therefore the transition from Winkler to Thompson out of Bruins training camp would best be described as a cut followed by a minor league assignment.
Supposedly (grain of salt) at about this same time, the Bruins made a competitive offer to the Americans for Roy Worters during his holdout. Americans didn't have a goalie during this time, even playing Normie Himes in net for a game. If it's true that Ross was in on a potential trade for Worters (grain of salt) that would suggest that he was shopping Winkler as the replacement for the Americans.
In any case, the fact that he went on to spend the rest of his career with three different minor league teams suggests that he was simply not able to make the cut as an NHL'er after 1928. His shutout record that season is extremely impressive, but it's probably more of an indicator of how good the Bruins were as a whole (3 finals in 4 years), plus the fact that leaguewide scoring in the two seasons in question was at the absolute lowest level in the history of the league.
As an example of how deceptive that record can be, George Hainsworth had a 1.05 GAA that year compared to Winkler's 1.51, across the same number of games. Yet despite allowing nearly 50% more goals, Winkler recorded 2 more shutouts than Hainsworth. In fact, Winkler also had a substantially higher GAA than Connell (1.24) who tied the shutout mark. The reason being that Winkler tended to get shelled (in 1928 terms) more often. Here's a breakdown of how the full-season goalies that season fared in terms of volume allowed:
Winkler
7 against - 1 (2%)
6 against - 1 (2%)
5 against - 1 (2%)
4 against - 2 (5%)
3 against - 7 (16%)
2 against - 6 (14%)
1 against - 11 (25%)
shutouts - 15 (34%)
Connell
7 against - 0
6 against - 1 (2%)
5 against - 0
4 against - 2 (5%)
3 against - 4 (9%)
2 against - 9 (20%)
1 against - 13 (30%)
shutouts - 15 (34%)
Hainsworth
7 against - 0
6 against - 0
5 against - 1 (2%)
4 against - 1 (2%)
3 against - 2 (5%)
2 against - 6 (14%)
1 against - 21 (48%)
shutouts - 13 (30%)
Worters
7 against - 0
6 against - 1 (2%)
5 against - 1 (2%)
4 against - 3 (7%)
3 against - 5 (11%)
2 against - 15 (34%)
1 against - 8 (18%)
shutouts - 11 (25%)
Benedict
7 against - 0
6 against - 0
5 against - 0
4 against - 4 (9%)
3 against - 9 (20%)
2 against - 9 (20%)
1 against - 16 (36%)
shutouts - 6 (14%)
Holmes
7 against - 0
6 against - 2 (5%)
5 against - 1 (2%)
4 against - 2 (5%)
3 against - 6 (14%)
2 against - 14 (32%)
1 against - 8 (18%)
shutouts - 11 (25%)
Chabot
7 against - 0
6 against - 1 (2%)
5 against - 0
4 against - 7 (16%)
3 against - 5 (9%)
2 against - 10 (23%)
1 against - 10 (23%)
shutouts - 11 (25%)
These numbers reveal that Winkler only comes out on top if we rank goalies specifically by shutouts. If instead we rank them by games with 0-or-1 goals against, we get:
Hainsworth: 34 (78%)
Connell: 28 (64%)
Winkler: 26 (59%)
Benedict: 22 (50%)
Chabot: 21 (48%)
Holmes: 19 (43%)
Worters: 19 (43%)
And if we count the games with 4-or-MORE goals against, we get:
Hainsworth: 2 (4%)
Connell: 3 (7%)
Benedict: 4 (9%)
Winkler: 5 (11%)
Worters: 5 (11%)
Holmes: 5 (12%)
Chabot: 7 (18%)
Of course, those types of games are virtually guaranteed Ws and Ls. But in an environment where the team scoring average is 1.90 goals per game, the critical question in winning hockey games is what proportion of games the team allows 2-or-fewer versus 3-or-more. These are the splits for this group:
Hainsworth - 40/4
Connell - 37/7
Worters - 34/10
Holmes - 33/11
Winkler - 32/12
Benedict - 31/13
Chabot - 31/13
What's particularly impressive about that last stat is that Winkler posted that ratio while playing for a Bruins team that won its division. That really uncovers how far the Bruins exceeded expectations relative to their rate of goals allowed -- even in spite of Winkler picking up the shutout record in the process!
Of course this is reading a lot into GA numbers, which is always a dubious approach. But when you put all of the information together -- the unimpressive GA spreads and ratios, the fact that the Bruins hastily identified a replacement and promoted him the following season, the fact that Winkler bounced around the minors after that -- the strong impression is that Winkler was not up to the standard of the team that surrounded him. Without getting in a time machine and actually watching him play hockey, the best I can do is draw a reasonable inference that Winkler was a "merely good" goalie who was in (literally) the best possible place at the best possible time to post an inflated shutout number.