Fix hash-table and package symcount

[Affonso-Gui]

In current implementation euslisp's hash tables only keep count of its active members.
This allows for the hash-table to end up cluttered by deleted members, causing problems such as #405.

The same behavior is seen in packages' internal and external vectors, which are implemented to act as hash tables.

(setq pkg (make-package "TEST"))
(dotimes (i 10000)
  (print i)
  (unintern (intern (symbol-pname (gensym)) pkg) pkg))
;; infinite loop

In this PR we solve this by keeping count of the used members (active + deleted) instead.

[Affonso-Gui]

I've updated the PR to use a different slot (fill-count) in hash-tables. This is done to keep the formatting and compatibility with the following ansi test:

(deftest hash-table-count.11
  (let ((table (make-hash-table)))
    (setf (gethash 'x table) 1)
    (values (hash-table-count table)
            (progn
              (remhash 'x table)
              (hash-table-count table))))
  1 0)

[Affonso-Gui]

ppc64le/debian:buster times out. Restarting..
Function compile.euslisp takes 47 min 42 sec
https://github.com/euslisp/EusLisp/runs/6304908324?check_suite_focus=true#step:3:2433

[k-okada]

confirmed that the test code works properly #510

2024-04-17T00:48:34.2745776Z   TEST-NUM: 0
2024-04-17T00:48:34.2746149Z     PASSED:   0
2024-04-17T00:48:34.2746523Z     FAILURE:  0
2024-04-17T00:48:34.2746980Z RESULT: test-hashtable-delete
2024-04-17T00:48:34.2747481Z   TEST-NUM: 1
2024-04-17T00:48:34.2747828Z     PASSED:   0
2024-04-17T00:48:34.2748247Z �[31m    FAILURE:  1�[0m
2024-04-17T00:48:34.2748739Z RESULT: test-hashtable-count
2024-04-17T00:48:34.2749174Z   TEST-NUM: 1
2024-04-17T00:48:34.2749396Z     PASSED:   1
2024-04-17T00:48:34.2749624Z     FAILURE:  0
2024-04-17T00:48:34.2749886Z RESULT: test-package-unintern
2024-04-17T00:48:34.2750206Z   TEST-NUM: 1
2024-04-17T00:48:34.2750518Z     PASSED:   0
2024-04-17T00:48:34.2750946Z �[31m    FAILURE:  1�[0m