I think that these "fish eggs" are the equivalent of the head tendril. While Koyukkuris develop from seed stage, they consume this egg to grow up; once they are grown enough, they break it and start to feed from mother's innards to complete their development.
I was thinking about how a mammalian pregnancy is carried on, considering the fact that during mammalian birth no bean paste comes out, unless the koyukkuri is enormous which results in a very messy birth because of the birth canal ripping, there must be some way that Koyukkuri develops from seed to baby.
I think in two possibilities for that: 1. The mamumamu cavity is flooded with the mother's filling, but that is not too likely to happen because seed Yukkuris don't have working mouths. 2. The mamumamu cavity develops one or more appendixes which resemble the forehead tendril, once Koyukkuris grow to baby stage, they fall from it and start to feed from mother's innards.
Comparing this picture with post #2226, I am noticing that the mamumamu is a cavity with two sections: the first section is the birth canal, something that resembles woman's vagina and cervix, and the second section is the womb, where Koyukkuris develop, which resembles woman's uterus.
The birth canal section is less tight than the womb section.