Solr indexing nested objects array - arrays

Question

We're trying to run an index in Solr (8.9.0 - Schemaless Mode) of a list of items that each contain 1 or 2 arrays of objects, with 1 or more records per array. The sample code below is the json we feed the index:
[
{
"id": 8270861,
"type": "Product",
"title": "Stripped T-shirt"
"tags": [{
"tagId": 218,
"tagIcon": "smile,happy",
"tagHelpText": "",
"tagValue": "grand"
},
{
"tagId": 219,
"tagIcon": "frown,sad",
"tagHelpText": "",
"tagValue": "grand"
}],
"keywords": [
{
"keywordId": 742,
"type": "color"
},
{
"keywordId": 743,
"type": "size"
}]
}
]
2 problems we run into:
PROBLEM 1:
The output of the solr query changes the format of the arrays to this (effectively removing the quotes):
...
"tags": [
"{tagIcon=smile,happy, tagHelpText=, tagId=218, tagValue=grand}",
"{tagIcon=frown,sad, tagHelpText=, tagId=219, tagValue=grand}"
],
"keywords": [
"{type=color, keywordId=742}",
"{type=size, keywordId=743}"
],
...
Is there a way to get the format of the arrays to come back the same way as fed into the index:
"tags": [
{ "tagId": 218, "tagIcon": "smile,happy", "tagHelpText": "", "tagValue": "grand" },
{ "tagId": 219, "tagIcon": "frown,sad", "tagHelpText": "", "tagValue": "grand"}
]
to avoid any conflicts when the value is a comma separated list. Are we missing some definition adjustments in the schema file? If so do we need to define the children of those parent keys(i.e. "tags.tagIcon")?
PROBLEM 2:
The index seems to reject an array with a single element. If we feed it the same json as above, but only one entry in the keywords array (or the tags array):
...
"keywords": [
{
"keywordId": 742,
"type": "color"
}]
...
it throws a code: 400 Unknown operation for the an atomic update: type"
Any suggestions on this would be welcome.