bug: filter financial data unusable in ownership_weight methodology

`current_shares_outstanding_all_classes` - a value that is used downstream to calculate `ownership_weight` (a value that also is derived from the `shares_all_classes`) - will cause divide by 0 problems downstream if it is calculated to be 0. It is unusable by the `ownership_weight` calculation if it is 0 (https://github.com/RMI-PACTA/pacta.portfolio.allocate/blob/4c96adb9856788d97293e41c0cd68b8951c3c56b/R/calculate_ownership_weight.R#L5).

There is already a filter in `prepare_financial_data` to remove this case, however it filters the wrong (but similar) column. 

An edge case was found by @Antoine-Lalechere in which ADR holdings with no similar EQ holdings pass through and yield a 0 value for `shares_all_classes`, causing the issue described above. 

This PR adjusts the filter to ensure that the actual required value is usable in the downstream analysis. 

See the reprex below for more information:
``` r
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(pacta.data.preparation)

# consider the case there an entity has an equity holding and an ADR holding
# consider further, that the equity holding that reflects the ADR holding has no price or share information
# note: this was a real case in the data
data <- tibble::tribble(
  ~fsym_id,          ~isin, ~factset_entity_id, ~adj_price, ~adj_shares_outstanding, ~issue_type, ~one_adr_eq,
  "ABCDE-F", "US123456789",         "12ABCD-E",      5.5,               12345678,        "AD",          NA,
  "ABCDE-G", "JP123456789",         "12ABCD-E",         NA,                      NA,        "EQ",          NA
)

issue_code_bridge <- tibble::tribble(
  ~issue_type_code, ~asset_type,
  "AD",    "Equity",
  "EQ",    "Equity"
) |> 
  pacta.data.preparation::standardize_asset_type_names() 

# our current methodology does NOT consider ADR holdings in the calculation of ownership_weight denominator
# what this means is that, in the case above, it would be impossible to calculate an appropriate ownership_weight
# since the denominator would be 0

# however, as can be seen below, the ADR holding does appear in the output, with 
# a value of 0 for current_shares_outstanding_all_classes. This is unusable downstream by the ownership_weight calculation
# 
# See: https://rmi-pacta.github.io/pacta.data.preparation/articles/share_ownership_methodology.html for more information
out <- pacta.data.preparation::prepare_financial_data(data, issue_code_bridge)

print("US123456789" %in% out$isin)
#> [1] TRUE
print(filter(out, isin == "US123456789")["current_shares_outstanding_all_classes"])
#> # A tibble: 1 × 1
#>   current_shares_outstanding_all_classes
#>                                    <dbl>
#> 1                                      0
```

<sup>Created on 2024-07-17 with [reprex v2.1.1](https://reprex.tidyverse.org)</sup>

R/prepare_financial_data.R

@@ -31,11 +31,12 @@ prepare_financial_data <-
         .by = "factset_entity_id"
       ) %>%
       filter(
+        # these filters are meant to drop financial data points that are functionally unusable for PACTA's ownership weight and fund expansion methodology
         case_when(
           asset_type == "Bonds" ~ TRUE,
           asset_type == "Others" ~ TRUE,
           asset_type == "Funds" ~ !is.na(.data$adj_price),
-          asset_type == "Equity" ~ .data$adj_price > 0 & .data$adj_shares_outstanding > 0
+          asset_type == "Equity" ~ .data$adj_price > 0 & .data$shares_all_classes > 0
         )
       ) %>%
       select(