You've got a Table for the data set, so make another Table (or just a range, but it looks like you like and/or are comfortable with Tables) for the list of fruits and the differentiator (column in the data set Table) to use for each. Perhaps you do that in E1:F3.
Then build the output table: not a Table, but just a regular range. Perhaps it will start in H1. Label that "Fruit", and move on to labelling the next ten, say, columns with this formula:
="Diff-"&(COLUMN()-8)
or whatever seems like good column headings for the differentiators for the fruits.
Why a normal range and not a Table? Because a Table will NOT allow a SPILL formula inside it to work. And you'd like to have that feature. (It is not because the column headings would convert to text... that wouldn't be any unhappy thing at all.)
In H2, place this formula for listing all the unique items in the data set's "Fruit" column:
=UNIQUE(INDIRECT("Table1["&H1&"]"))
or, you know, don't connect it like that to H1, just use the direct:
=UNIQUE(Table1[Fruit])
So now you have suitable headings and a column collecting all the fruit names that exist in the data set "Fruit" column... the set of unique names, not all the entries. All you need now is their unique existing differentiators.
=TRANSPOSE(UNIQUE(INDEX(FILTER(Table1[[Color]:[Size]],Table1[Fruit]=H2),,MATCH(XLOOKUP(H2,Table2[Fruit],Table2[Differentiator]),Table1[[#Headers],[Color]:[Size]],0))))
XLOOKUP(H2,Table2[Fruit],Table2[Differentiator]) uses the value in column H to lookup the fruit's differentiator in the second Table. It passes that value to the MATCH() function wrapping it which compares it to the headers of the data set Table and returns a number that you will use as a column number in the eventual INDEX() function.
For that INDEX() function you want all the otherwise acceptable rows to be used hence the empty commas that would usually have a row number between them.
INDEX() needs a data set to work with and that's what FILTER() does for you. Its test looks for the fruit in H2 in the data set Table and returns those rows that match the fruit. But it returns all the characteristics columns (two here, but you could expand it to more someday).
INDEX() then uses the column number found by MATCH() to return only one characteristics column to the formula.
Now that you have only the right column and only the rows in it that are for the fruit in H2, you can put UNIQUE() to work.
It will return a column of the unique values you need. But you need a row, so the TRANSPOSE() function finishes things off for you.
SPILL fills the row for you. Copy down as required.
Then collect the output data in whatever way you need (copy and paste in some other program ("Shopify" directly, perhaps), paste into a transfer spreadsheet you save as a CSV file, or perhaps a spreadsheet works, however you go about it. SPILL arrays copy and paste nicely.
Then when you have time, dress it up with LET(), giving the horrid "Table1[[Color]" references real readable names, and placing the input that you might wish to edit, like "H2" at the absolute start so they are easy to find and edit.
Finally, when setting up the T/tables, consider separating them, perhaps onto different sheets, to allow easy expansion of them as you add more kinds of items over time.
